Antiangiogenic therapy has become the most important treatment modality for patients with advanced hepatocellular carcinoma (HCC). In this study, the authors investigated levels of alpha-fetoprotein (AFP) as a potential biomarker for treatment efficacy of antiangiogenic therapy.
Patients with advanced HCC who had been enrolled in 3 prospective phase 2 clinical trials that evaluated either sorafenib, bevacizumab, or thalidomide in combination with a potentially antiangiogenic, metronomic, oral 5-fluoropyrimidine as first-line systemic therapy were included. An early AFP response was defined as a decline >20% from baseline after 2 to 4 weeks of treatment. AFP response was analyzed for its association with treatment efficacy and survival outcome.
Seventy-two patients were included for early AFP response evaluation, and 12 of those patients (17%) were classified as early AFP responders. Early AFP responders, compared with nonresponders, had a significantly improved overall response rate (33% vs 8%; P = .037) and a significantly improved disease control rate (83% vs 35%; P = .002), which was defined as the percentage of patients who had an objective response plus stable disease for a minimum of 8 weeks. AFP responders, compared with nonresponders, also had longer median progression-free survival (PFS) (7.5 months vs 1.9 months; P = .001) and longer median overall survival (OS) (15.3 months vs 4.1 months; P = .019). In a multivariate analysis, AFP response remained a significant independent predictor of better PFS and OS.
The treatment of advanced hepatocellular carcinoma (HCC) that is not amenable to curative or locoregional therapies remains a great challenge in clinical practice. In general, cytotoxic chemotherapy is ineffective in the treatment of advanced HCC.1-4 In the Asia-Pacific region, the median overall survival (OS) o patients with advanced HCC patients ranges from 2 months to 4 months with supportive care.5-7
Sorafenib is a multitargeted agent that has antiangiogenic activity.8 In 2 phase 3, placebo-controlled, randomized trials, sorafenib significantly improved the time-to-tumor progression (TTP) and OS in patients with advanced HCC.6, 9 Several other antiangiogenic agents, such as sunitinib, bevacizumab, thalidomide, brivanib, and ABT-869, also have demonstrated variable clinical activities in advanced HCC.10-15 However, antitumor activities of antiangiogenic agents remain modest, and the objective tumor response rates produced range from 3% to 13% assessed by conventional response criteria.6, 9-15 Therefore, it is imperative to search for biomarkers for antiangiogenic agents in HCC.
Alpha-fetoprotein (AFP) is a glycoprotein that is expressed by HCC and is secreted into the serum of approximately 70% of patients with HCC. AFP levels have been widely monitored in the management of HCC and provide 1 criterion among the HCC diagnostic criteria in the American Association for the Study of Liver Diseases guidelines.16 Previous studies have demonstrated that the baseline AFP level is a significant prognostic factor for various stages of HCC.16-22 Recently, Chan and colleagues reported that a decline in the post-treatment AFP level was associated with radiologic tumor response and survival in patients with advanced HCC who received doxorubicin-based combination chemotherapy.23 Vora et al. also reported that post-treatment AFP changes were closely associated with clinical outcomes in patients who received various types of systemic therapy.24 Whether changes in AFP can predict the clinical outcomes of patients with advanced HCC who receive antiangiogenic therapies remains unclear. In the current study, we analyzed patients with advanced HCC who had been received antiangiogenic agents in 3 prospective phase 2 trials, and we retrospectively studied the association of AFP response with tumor response and survival outcomes.
MATERIALS AND METHODS
The study population consisted of patients with HCC who were enrolled prospectively in 3 phase 2 studies of first-line systemic therapy for advanced diseases at National Taiwan University Hospital (NTUH), Taipei, Taiwan, between 2005 and 2008.25-27 To be enrolled in these phase 2 studies, these patients needed to have metastatic or locally advanced HCC that was not amenable to locoregional therapies, including surgery, transcatheter arterial (chemo)embolization, and local ablation.
The 3 phase 2 trials tested the hypothesis that the antitumor effect of antiangiogenic therapy in patients with advanced HCC could be enhanced by metronomic chemotherapy, which was defined as the uninterrupted administration of chemotherapeutic agents at lower doses for a prolonged time.28 In preclinical models, metronomic chemotherapy reportedly was antiangiogenic rather than cancer cell cytotoxic.29-32 The detailed treatment regimens and eligibility criteria of the 3 trials are provided in Table 1. Therapy continued until tumor progression or until intolerable toxicities occurred.
Table 1. Regimens and Key Eligibility Criteria for the Phase 2 Trials Analyzed in the Current Studya
Defined as the presence of hepatic tumor(s) with image findings compatible with HCC and a persistent elevation of serum alpha-fetoprotein level ≥400 ng/mL in patients with chronic hepatitis B or C virus infection.
The main eligibility criteria of the 3 trials were similar. For example, patients had to provide written informed consent; and they had to have at least 1 measurable lesion according to Response Evaluation Criteria in Solid Tumors (RECIST) criteria (version 1.0),33 adequate liver reserve (Child-Pugh Class A), and adequate major organ functions. The other key eligibility criteria are listed in Table 1.
Patients had imaging studies obtained for tumor response evaluation every 6 to 8 weeks. The efficacy endpoints of the studies included objective tumor response according to RECIST criteria; the disease control rate, which was defined as the percentage of patients who had an objective tumor response or disease stabilization for a minimum of 8 weeks; progression-free survival (PFS); and OS.
The 3 phase 2 clinical studies had been approved by the Institute Research Ethical Committee of NTUH. The recruitment notification has been posted online (available at: http://www.clinicaltrials.gov accessed on December 22, 2009).
Examination of AFP Level and Definition of Early AFP Response
According to the protocol designs, serum AFP levels were measured at baseline and every 2 or 3 weeks after the start of treatment until patients developed tumor progression or discontinued treatment. Serum AFP levels were measured by using a microparticle enzyme immunoassay (Abbott Laboratories, Chicago, Ill) during the prospective trials by the Central Laboratory of Clinical Biochemistry in the Department of Laboratory Medicine at NTUH.
Patients who had normal baseline AFP levels (<20 ng/mL) were excluded in the AFP response evaluation because, they are less likely to have a significant decline in AFP level. Early AFP response was defined as a decline >20% from the baseline AFP level within the first 4 weeks of initiating treatment.23 If more than 1 AFP level was available during the first 4 weeks after treatment, then the higher level was chosen.
Statistical analyses were performed using the SAS statistical software package (version 9.1.3; SAS Institute, Inc., Cary, NC). In statistical testing, 2-sided P values ≤.05 were considered statistically significant. The associations between treatment response or disease control with AFP response were examined by chi-square tests or Fisher exact tests.
The Kaplan-Meier method was used to estimate survival. Early AFP response, along with other clinicopathologic variables, was evaluated first in univariate analyses and then in multivariate analyses using a Cox proportional hazards model to determine potential associations with survival outcome in patients with HCC. The clinicopathologic variables included treatment cohort, age, sex, performance status, hepatitis etiology, vascular invasion, extrahepatic involvement, metastatic sites, Cancer of the Liver Italian Program (CLIP) score,34 Okuda stage, and Barcelona Clinic Liver Cancer (BCLC) stage. In the stepwise variable selection procedure during multivariate analysis, all significant and nonsignificant covariates were considered, and the significance levels for entry and for stay were set at ≥0.15.
The 3 phase 2 trials enrolled a total of 141 patients, and 107 of those patients were from NTUH. Eighty-five patients (79%) had elevated baseline AFP levels. Patients who had no post-treatment AFP evaluation because of rapid disease progression (11 patients) or missing follow-up AFP data (2 patients) were excluded; thus, 72 patients were analyzed for early AFP response.
Among the 72 patients, 90% were men, the median age was 54 years (range, 24-83 years), 78% were seropositive for hepatitis B virus surface antigen (HBsAg), 97% had either extrahepatic metastasis (64%) or macrovascular invasion (61%), and 99% were classified with BCLC stage C disease. Their demographic features are summarized in Table 2. With regard to treatment response, there were 0 complete responses and 9 partial responses, for an overall tumor response rate of 13%. The median PFS was 2 months, and the OS was 4.8 months for all 72. Survival outcomes did not differ significantly between patients who received different regimens (log-rank test: PFS, P = .200; OS, P = .186).
Table 2. Patient Characteristics
No. of Patients
HBsAg indicates hepatitis B surface antigen; HCV, hepatitis C virus; AFP, alpha-fetoprotein; ECOG PS, Eastern Cooperative Oncology Group performance status; BCLC, Barcelona Clinic Liver Cancer; CLIP, Cancer of the Liver Italian Program.
Median age (range), y
Sites of disease
Abdominal lymph nodes
Extrahepatic metastasis or macroscopic vascular invasion
Macroscopic vascular invasion
Mean AFP level (range), ng/mL
Early AFP Response and Treatment Outcomes
Twelve of the 72 patients (17%) were classified as early AFP responders. Compared with nonresponders, early AFP responders had a significantly higher objective response rate (ORR) (33% vs 8%; P = .037) and disease control rate (83% vs 35%; P = .002) (Table 3).
Table 3. Association of Early Alpha-Fetoprotein Response With Treatment Efficacy
Treatment response included patients who achieved complete and partial responses; disease control included patients who had a treatment response and stable disease.
Two-sided Fisher exact test.
Two-sided chi square test.
Early AFP responders
Early AFP nonresponders
Early AFP response was a favorable predictor of PFS and OS. Early AFP responders, compared with nonresponders, had a significantly longer median PFS (7.5 months vs 1.9 months; P = .001) (Fig. 1, top) and median OS (15.3 months vs 4.1 months; P = .019) (Fig. 1, bottom).
Independent Prognostic Role of Early AFP Response
A multivariate analysis was conducted to elucidate the independent prognostic significance of early AFP response and other clinical parameters (Table 4). An early AFP response (hazard ratio [HR], 0.307; P = .003) was an independent factor associated with longer PFS. Other independent factors that had a significant impact on PFS included being a man, positive serum HBsAg status, poorer performance status, and higher CLIP score, and every variable was associated significantly with shorter PFS. When we analyzed for the impact on OS, an early AFP response (HR, 0.356; P = .017) and a lower CLIP score were significant independent factors associated with longer OS.
Table 4. Cox Proportional Hazards Model for Predictors of Progression-Free and Overall Survival
HR indicates hazard ratio; CI, confidence interval; HBsAg, hepatitis B surface antigen; ECOG PS, Eastern Cooperative Oncology Group performance status; CLIP, Cancer of the Liver Italian Program; AFP, alpha-fetoprotein.
ECOG PS 1 (vs 0)
CLIP score 2-4 (vs 0-1)
Early AFP response
CLIP 2-4 (vs 0-1)
Early AFP response
In the current study, we demonstrated that an early AFP response, ie, a decline >20% from the baseline AFP level within the first month after the initiation of treatment, was associated significantly with better efficacy endpoints of antiangiogenic therapies in patients with advanced HCC. The early AFP responders in this study had a significantly better treatment response, better disease control, longer PFS, and longer OS compared with AFP nonresponders. It is noteworthy that the AFP nonresponders had a disappointingly low disease control rate (35%) and extremely poor median PFS and OS (1.9 months and 4.1 months, respectively).
Unlike the baseline AFP level, the prognostic value of which has been explored extensively in various stages of HCC, AFP changes during treatment have been evaluated less frequently for their predictive or prognostic significance. Three recent studies in patients with advanced HCC who received various types of systemic therapy, such as thalidomide,35 doxorubicin-based combination chemotherapy,23 and chemotherapy with or without molecularly targeted therapies,24 concluded that an AFP change during treatment is a useful surrogate for predicting tumor response and patient survival, a finding that was supported by the current report focusing on antiangiogenic therapy as the treatment strategy.
However, previous studies have not agreed on the definition of AFP response with regard to the magnitude of AFP decline and the time points for evaluating AFP levels.23, 24, 35 The AFP decline is either a 50% reduction24, 35 or a 20% reduction from baseline23; and the post-treatment time points were 4 weeks to 8 weeks, 7 weeks to 9 weeks, or different time points throughout the course of treatment.23, 24, 35 The early AFP response that we used in the current study was based on AFP changes within the first month after the initiation of treatment. To our knowledge, this is the earliest time point ever used in similar studies, but it appears to be the most relevant for the clinical practice of treating patients with advanced HCC. In the Asia-Pacific region, the median PFS and OS of patients with advanced HCC can be as short as 2 months to 4 months.5-7 A biomarker that can be detected in the early phase of treatment is valuable, because it may help identify a subgroup of patients who have a poor prognosis and are candidates for novel or other treatment strategies.
We adopted a 20% decline from the baseline AFP level as an AFP response, because the criterion of a 50% reduction at such an early time point may be so stringent that it can exclude a substantial portion of patients who may obtain a clinical benefit without having a rapid and dramatic tumor response. Indeed, when we performed the analysis using a 50% reduction from baseline as an AFP response, much fewer patients (only 4) were classified as responders. Despite this limitation, the responders under such a stringent definition still had highly favorable efficacy endpoints. Whether a greater AFP decline confers better therapeutic efficacy is an interesting hypothesis that is worth testing. However, we did not observe such an association in the current study. For example, patients who had a 50% reduction in AFP level, compared with patients who had a 20% to 50% reduction, did not have a higher ORR, disease control rate (Fisher exact test; P = .222 and P = .470, respectively) or longer median PFS and OS (log-rank test; P = .800 and P = .487, respectively).
A negative predictor of poor response or no response to antiangiogenic therapy may be as important as a positive prognostic or predictive marker. A recent example of K-ras mutation in patients with advanced colorectal cancer as a negative predictor for the efficacy of cetuximab, a monoclonal antibody that targets the epidermal growth factor receptor, demonstrated how a negative predictor may help to tailor better, personalized cancer therapy.36 In terms of our current results, it was reasonable to suspect that patients who had a rise in AFP level after treatment would have poorer treatment response and survival outcomes. However, the association of poor treatment outcomes, including nonresponse or independently worse predictors for PFS and OS, with “early AFP progression,” ie, elevated AFP levels within the first month after treatment, could not be identified in our cohort (data not shown). Future studies in larger patient cohorts and/or with more stringent criteria for “early AFP progression” may be helpful to verify the usefulness of such an approach.
Biomarkers for predicting the efficacy of antiangiogenic therapy have been investigated actively, and the potential surrogate markers include circulating angiogenesis-related factors like vascular endothelial growth factor (VEGF), functional imaging studies like dynamic contrast-enhanced magnetic resonance image (DCE-MRI), and the enumeration of circulating endothelial cells (CECs) and circulating endothelial progenitors cells (CEPs).37 Serum or plasma levels of VEGF or other angiogenic factors that can be determined readily with established procedures have not been effective predictors in patients with advanced HCC who receive antiangiogenic-targeted agents.11, 15 DCE-MRI and the enumeration of CECs/CEPs, which are scientifically sound, are not readily applicable, because the standard protocols for these approaches remain undetermined, and the required instruments and procedures are restricted mainly to academic research centers. In contrast, the measurement of serum AFP has been a routine clinical biochemistry test for decades, and AFP monitoring can be implemented immediately, easily, and broadly in the clinic as a potential surrogate for the efficacy of antiangiogenic therapy in patients with advanced HCC.
Potential limitations of our study included a relatively small patient number and a less homogenous patient population, which consisted of 3 phase 2 study cohorts. However, the 3 different combination therapies were based on the same rationale and similar designs, ie, the combination of an agent with antiangiogenic activity, such as sorafenib, and metronomic chemotherapy using an oral fluoropyrimidine. Furthermore, we evaluated treatment cohort as a variable in the multivariate analysis and observed no interactions between treatment cohort and the predictive value of AFP level.
In conclusion, our current results indicate that an early AFP response is a significant predictor of efficacy and survival outcomes for patients with advanced HCC who receive antiangiogenic therapies. The evaluation of AFP levels, especially in the early phase of treatment, warrants further exploration in future clinical studies.
CONFLICT OF INTEREST DISCLOSURES
This work was supported by a grant from the Department of Health, Executive Yuan of Taiwan to the National Center of Excellence for Clinical Trial and Research, National Taiwan University Hospital (grant DOH96-TD-B-111-001) and by grant NSC98-3112-B-002-038.