Bevacizumab-induced Hypovascular Hepatocellular Carcinoma Treated by Transarterial Chemoembolization in a Patient with Metastatic Colon Cancer

The angiogenesis of hepatocellular carcinoma (HCC) is mediated by growth factors, particularly vascular endothelial growth factor (1). Serum vascular endothelial growth factor levels in patients with HCC correlate with malignant behavior and clinical outcomes (2, 3). Early clinical studies have shown that the anti–vascular endothelial growth factor antibody bevacizumab might be able to deregulate angiogenesis in HCC (4). Angiogenesis inhibited by bevacizumab appears as a significant reduction in tumor enhancement on computed tomography (CT) perfusion (5) and dynamic contrast-enhanced magnetic resonance images (4). Here, we treated a patient with histopathologically proven HCC and metastatic colon cancer by transarterial chemoembolization before and after chemotherapy with bevacizumab.

An 80-year-old man with hepatitis C–related liver cirrhosis was admitted to undergo treatment for colon cancer with multiple liver metastases. Dynamic contrast-enhanced CT showed early-enhanced tumors at hepatic segments III and VI accompanied by multiple metastatic colon tumors. Needle biopsy specimens of both tumors obtained during surgical resection of the transverse colon cancer were histopathologically diagnosed as well differentiated HCC. Chemoembolization proceeded 41 days after surgery by coaxially positioning a 2.4-F microcatheter (Sniper 2; Clinical Supply, Gifu, Japan) through a 4-F catheter via the femoral artery. Selective appropriate hepatic angiography revealed well enhanced tumors at the left hepatic lobe (Figure, a). Therefore, the left hepatic artery and the right posterior arterial segment were separately embolized with gelatin particles (Gelpart; Astellas Pharma, Tokyo, Japan) after infusion with a 5-mL emulsion of iodized oil and 40 mg of epirubicin hydrochloride at a ratio of 2:1. Follow-up CT at 7 days after chemoembolization confirmed sufficient iodized oil accumulation in both lesions.

View Large Image Download to PowerPoint

Figure. Comparative images from an 80-year-old man with HCC and metastatic colon cancer obtained before (a), during (b,c), and after terminating (d) bevacizumab therapy. (a) Selective appropriate hepatic angiogram shows HCC as well defined hypervascular tumor at lateral hepatic segment (arrow). (b) Arterial phase image of dynamic contrast-enhanced CT obtained after bevacizumab therapy for 6 months shows ill-defined, hypodense, recurrent HCC located around previously accumulated iodized oil at lateral hepatic segment (arrow). Similar hypovascular tumors at the right hepatic lobe are colon cancer metastases. (c) Selective common hepatic angiogram shows recurrent HCC as less hypervascular tumor with obscure margin at left hepatic lobe (arrow). Lesion vasculature seems narrower and sparser versus before bevacizumab therapy. (d) Selective common hepatic angiogram after terminating bevacizumab therapy for 3 months shows restored vascularity of recurrent tumor (arrow).

First-line systemic chemotherapy for the metastatic colon cancer comprised 12 cycles of oxaliplatin (85 mg/m2), 5-flourouracil (400 mg/m2), levofolinate (200 mg/m2) and 5-fluorouracil (2,400 mg/m2; 46-hour infusion; ie, mFOLFOX6 regimen) without bevacizumab every 2–4 weeks. Recurrent HCC was not obvious during the course of first-line chemotherapy. Second-line chemotherapy comprised six cycles of bevacizumab (10 mg/kg) combined with irinotecan (180 mg/m2), levofolinate (200 mg/m2), and 5-fluorouracil (2,400 mg/m2; 46-hour infusion; ie, FOLFIRI regimen) every 2–4 weeks depending on adverse effects. Two months after bevacizumab administration, dynamic CT revealed a hypodense lesion around the previously accumulated iodized oil at the left hepatic lobe. Although that lesion was regarded as metastatic colon cancer at the time, the hypodense lesion increased to approximately 5 cm in diameter within 6 months of bevacizumab administration (Figure, b). The lesion could not be discriminated from coexisting metastatic colon cancer, but a percutaneous biopsy specimen was histopathologically diagnosed as well differentiated HCC compatible with the specimen previously obtained during surgery. During the second chemoembolization session, common hepatic angiography demonstrated a less hypervascular tumor at the left hepatic lobe (Figure, c). Because CT images (Figure, b) and angiograms had previously shown similar, small, hypodense tumors in the right hepatic lobe, the possibility that small hypovascular HCC mimicked metastatic colon cancer in the right lobe could not be excluded. Therefore, we separately embolized the left and right hepatic arteries with gelatin particles after infusion with a chemotherapeutic emulsion of 6 mL of iodized oil and 50 mg of epirubicin hydrochloride at a ratio of 2:1. Follow-up CT revealed irregular and insufficient iodized oil accumulation within the lesion on the left lobe, and none in any of the lesions in the right lobe. Because the inadequate accumulation of iodized oil by the target HCC might be attributable to bevacizumab-induced angiogenesis inhibition, we excluded bevacizumab from second-line chemotherapy, but continued with other chemotherapeutic agents. Three months after bevacizumab withdrawal, dynamic CT revealed a recurrent tumor with restored tumor vascularity. Angiography also delineated the recurrent lesion as a more hypervascular tumor (Figure, d). Therefore, we repeated the chemoembolization for the recurrent lesion, and follow-up CT confirmed sufficient iodized oil accumulation within the target lesion. The HCC has not locally recurred during a follow-up period of 12 months.

Hypovascular HCC can be difficult to accurately diagnose by imaging alone. Indeed, we could not discriminate hypovascular recurrent HCC from coexisting metastatic colon cancer on dynamic CT images. Digital subtraction angiography demonstrated recurrent HCC with bevacizumab treatment as tumors that were relatively more hypervascular than metastatic colon tumors, but vascularity was obviously decreased compared with the lesions before bevacizumab therapy. Bevacizumab probably reduced HCC vascularity, resulting in inadequate iodized oil accumulation within the target HCC. Meanwhile, the vascularity of the HCC was almost completely restored after discontinuing bevacizumab for 3 months. Mancuso et al (6) showed that the vascular endothelial growth factor receptor inhibitor axitinib causes a loss of 50%–60% of tumor vasculature in 7 days, and that revascularization is fully restored to islet cell tumors in mice within 7 days of drug withdrawal. However, the revascularization process of HCC after discontinuation of bevacizumab is unknown. We did not perform dynamic contrast-enhanced CT until 3 months after bevacizumab withdrawal, so exactly how and when tumor vascularity recovered in our patient could not be determined. However, discontinuing bevacizumab for 3 months in our patient restored vascularity to the HCC, thereby preventing an inadequate outcome of chemoembolization.

The outcomes of chemoembolization for HCC on a background of bevacizumab therapy were insufficient in this patient. However, whether outcomes actually correlate with the treatment response and clinical effectiveness requires further investigation because bevacizumab itself confers a potential benefit by inhibiting neovascularization in tumors, which prevents their growth (7). Continuing to administer bevacizumab in this patient without repeating chemoembolization would have ultimately extended survival just as well as stopping bevacizumab and performing chemoembolization. In addition, the possibility of spontaneous or chemoembolization-induced differentiation of the hypervascular HCC into lower-grade malignancy could not be excluded even though a pathologic examination confirmed the coincidence of both lesions before and after bevacizumab therapy. We also could not eliminate the effects of other chemotherapeutic agents administered to treat the metastatic colon cancer on the neovascularization process of HCC. Further investigation is needed to evaluate the effectiveness of chemoembolization for HCC in patients receiving bevacizumab.