Combined sorafenib and yttrium-90 radioembolization for the treatment of advanced hepatocellular carcinoma

Background and Aims In this pilot study, we assessed the safety and tolerability of combining sorafenib with 90 Y radioembolization for the treatment of unresectable hepatocellular carcinoma ( hcc ). Methods The study, conducted prospectively during 2009–2012, included eligible patients with unresectable hcc and a life expectancy of at least 12 weeks. Each patient received sorafenib (400 mg twice daily) for 6–8 weeks before 90 Y treatment. Safety and tolerability were assessed. Results Of the 40 patients enrolled, 29 completed treatment (combined therapy). In the initial cohort, the most common cause of hcc was hepatitis C (32.5%), and most patients were staged Child A (82.5%). The 29 patients who completed the study had similar baseline characteristics. Grades 1 and 2 toxicities accounted for 77.8% of all adverse events reported. The most common toxicities reported were fatigue (19.0%), alteration in liver function (7.9%), and diarrhea (6.3%). There were 12 grade 3 and 2 grade 4 toxicity events reported. One patient died of liver failure within 30 days after treatment. During the study, the sorafenib dose was reduced in 6 patients (20.7%), and sorafenib had to be interrupted in 4 patients (13.8%) and discontinued in 4 patients (13.8%). The disease control rate was 72.4% per the modified Response Evaluation Criteria in Solid Tumors, and tumour necrosis was observed in 82.8% of patients. Overall survival in patients undergoing combined therapy was 12.4 months. Conclusions Preliminary results demonstrate the safety and tolerability of combining 90 Y radioembolization and sorafenib for advanced hcc . A larger prospective study is needed to determine the extent of the survival benefit.


INTRODUCTION
Hepatocellular carcinoma (hcc) is the most common form of primary liver cancer, the 5th most common cancer in adult men worldwide, and the 3rd leading cause of cancerrelated death globally [1][2][3][4] . The incidence of primary liver cancer in the Western world is increasing. In the United States, a 25% increase in hcc cases was observed between 1994 and 2002 5 . Moreover, despite a general decline in the cancer mortality rate for most sites, mortality for hcc is increasing, and hcc is the fastest-growing cause of cancerrelated death in men 6 . Curative treatments, including liver transplantation and liver resection, carry a 5-year survival of 50% -75%, but can be used in only 30% -40% of the cases in Europe and United States 7,8 . In patients for whom resection or transplantation is not readily feasible, locoregional treatments are offered: radiofrequency, cryo-, percutaneous alcohol injection, or microwave ablation 9,10 and bland embolization, radioembolization, or transarterial chemoembolization, which might be conventional or might use drug-eluting beads 11,12 . Designing new treatments has been a challenge, given that most patients present with cirrhosis and associated limitations in the synthetic hepatic function. In patients with more advanced stages of hcc, best supportive care

Sorafenib Therapy
In 2008, the multicentre phase iii double-blind randomized placebo-controlled sharp clinical trial 14 , conducted in patients with advanced hcc, demonstrated that, compared with placebo, sorafenib conferred a survival advantage of 3 months (hazard ratio: 0.69; p = 0.00058). Median overall survival was 10.7 months in the sorafenib arm compared with 7.9 months with placebo. Time to tumour progression was also increased by 2.7 months (range: 2.8-5.5 months) with sorafenib. That same year, sorafenib became the first approved systemic treatment for patients with advanced hcc.
Sorafenib is an oral multi-kinase inhibitor with demonstrated antiangiogenic and antiproliferative effects. It acts by inhibiting the serine/threonine kinases Raf-1 and B-Raf and the receptor tyrosine kinase activity of vascular endothelial growth factor receptors 1, 2, and 3, and platelet-derived growth factor receptor 16 -pathways that have both been implicated in the pathogenesis of hcc 16,17 . Sorafenib is now the only systemic treatment shown to improve outcomes in advanced hcc.

Y Therapy
Since its first use in 1969, 90 Y therapy has evolved into two commercially available radioembolization devices: glass microspheres called TheraSphere (BTG International, London, U.K.) and resin microspheres called SIR-Spheres (Sirtex Medical, Sydney, Australia). TheraSphere microspheres 18 are intra-arterial, minimally embolic, 20-μm to 30-μm insoluble glass particles loaded with 90 Y. Compared with the resin microspheres, they have higher specific activity (2500 Bq) and a lower number of spheres per dose (1.2 million/3 GBq) 19 , allowing for delivery of the radiotherapeutic agent with less secondary embolization effect.
Unlike external-beam radiation, radioembolization beads are introduced through a selected hepatic artery and are consequently trapped in the microvasculature because of the size differential between the beads and the arteriolar-capillary diameter 20 . Unlike a healthy liver, which relies on the portal vein, intrahepatic malignancies rely on the hepatic artery for their blood supply [21][22][23] , thereby providing an effective and relatively safe avenue for delivering high doses of radioembolization agent while avoiding radiation-induced liver disease. Microspheres with 90 Y exert their local radiotherapeutic effect primarily from beta-particle emission, which has an average therapeutic range of 2.5 mm in tissue (0.94 MeV per beta particle), with a maximal range of less than 1 cm. The isotope has a physical half-life of 64.2 hours and decays to stable 90 Zr 18,24,25 .
In recent years, 90 Y radioembolization has been investigated in numerous cancer types and has shown promising results in patients with hcc [26][27][28][29][30][31][32][33][34][35][36] . The 2008 meta-analysis of 90 Y studies by Vente et al. 37 showed a 78% response from treatment with glass microspheres in patients with hcc. Ongoing research is encouraging, showing a partial response in up to 55%-79% of patients 30,33 , with significant delays in tumour progression and vascular invasion, low risk for microembolic effect, and effectiveness in patients with limited hepatic reserve 17 .

Rationale for Combining Sorafenib and 90 Y Therapy
Above and beyond sorafenib's ability to increase time to progression and overall survival in patients with very advanced hcc, and 90 Y therapy's ability to increase treatment response rates in these patients, there is theoretical rationale for combining the two treatments-specifically using 90 Y. Encouraging results from radioembolization and sorafenib studies also support a rationale for the combination. Tumour vasculature is tortuous and leaky 38,39 , and although it can deliver oxygen and nutrients to the core of the tumour, it is thought to be inefficient compared with normal vessels. Antiangiogenic agents have been shown to normalize the vessels and allow them to deliver oxygen more efficiently to the core of hcc tumours 40,41 , which in fact has been associated with enhanced radiation-induced tumour regression because sensitivity to radiation is increased 38 . The synergy between intra-arterial radioembolization and sorafenib therapy has also been investigated in an in vitro study, which showed an advantage of combining the two treatment modalities in human and murine hepatoma cell lines 42 . The same observations were also made in studies using in vivo models [43][44][45] .
We hypothesized that treating patients with sorafenib before radioembolization could increase tumour sensitivity to radiation by normalizing the tumour vasculature, thus increasing oxygenation of the tumour. The aim of the present pilot study was to evaluate the safety and tolerability of combining sorafenib with 90 Y radioembolization in the treatment of unresectable hcc and to evaluate disease progression and overall survival.

Patients
The target population consisted of consecutive eligible patients more than 18 years of age with advanced hcc as defined by guidelines for cirrhotic subjects from the American Association for the Study of Liver Diseases 41 .
Eligible patients required an ecog (Eastern Cooperative Oncology Group) performance status of 2 or less, a Child-Pugh liver function score of A-B, and a life expectancy of 12 weeks or more. Adequate renal (serum creatinine < 1.5 the upper limit of normal), hematologic (in particular, platelets > 50×10 9 /L), and cardiac function were also required; a dedicated cardiac work-up, including echocardiography was ordered only if patients had a history of cardiovascular disease or congestive heart failure. Patients who had previously received locoregional, surgical, or systemic cancer treatment were not excluded. Consecutive eligible patients attending the McGill University Health Centre hcc clinic were enrolled from April 2009 to August 2012. Patients were excluded if they were scored Child-Pugh C; had an ecog status of 3-4; had extensive extrahepatic disease or presented with liver tumours representing more than 50% of liver volume; had significant pulmonary disease (chronic obstructive pulmonary disease), cardiac disease (New York Heart Association class i, active coronary artery disease, and uncontrolled hypertension), active infection or hiv infection, recent gastrointestinal bleeding (<30 days), organ allograft, or major surgery within 4 weeks of entry; or had received biologic response modifiers such as granulocyte colony-stimulating factor 3 weeks before entry or had undergone autologous bone-marrow transplant. All patients provided informed consent, and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by our institution's human research committee (protocols SDR-08-041 and BAY-14270).
All patients underwent a pre-treatment evaluation that included medical history, ecog evaluation, laboratory work-up [alpha-fetoprotein (afp), virology, renal and liver function tests] and imaging ( Figure 1). Patients were classified according to the TNM system for hcc and the Barcelona Clinic Liver Cancer algorithm. Per standard protocol, all patients underwent combined diagnostic computed tomography (ct)-magnetic resonance imaging, hepatic arterial ct angiography, and mesenteric angiography before 90 Y treatment.

Treatment
All patients began oral sorafenib [initial dose: 400 mg twice daily (200-mg tablets)] and then underwent weekly liver function testing, hematology, standard biochemistry, and afp measurement, as well as biweekly medical consultation for a period of at least 6-8 weeks before 90 Y treatment. During that period, patients also underwent planning for 90 Y radioembolization, which included mesenteric angiography to map the hepatic vascular anatomy and arterial supply to the tumour. Coil embolization of non-target vessels was performed to allow for safe infusion of 90 Y microspheres at the time of the actual treatment. To detect significant shunting that would cause delivery of the microspheres to the lungs and gastrointestinal tract, 99m Tc macro-aggregated albumin was administered intraarterially to measure shunt fraction.
After the 6-to 8-week sorafenib regimen, standardtechnique 90 Y radioembolization was performed as a 1-day outpatient procedure under conscious sedation in the interventional radiology suite of the Royal Victoria Hospital ( Figure 1). Sorafenib was not interrupted on the day of the 90 Y treatment. The 90 Y-containing microspheres were delivered to the tumour by catheterization of the artery supplying the tumour. Radiation dose was calculated according to the manufacturer's guidelines, based on volume data from ct imaging of the liver. If a patient had bilobar disease, a second 90 Y radioembolization to treat the other side of the liver was performed approximately 3 weeks after the first treatment.
Sorafenib treatment interruptions and dose reductions for drug-related adverse effects followed the product monograph for sorafenib and the procedures in the sharp trial 14 . After 90 Y radioembolization, treatment with oral sorafenib (400 mg twice daily) continued for as long as patients found it tolerable, as assessed by the treating physician using the Common Terminology Criteria for Adverse Events, version 4.0.
After 90 Y radioembolization, patients underwent biweekly liver function testing, hematology, standard biochemistry, and afp measurement, as well as monthly medical consultation. After 3 months of follow-up and based on health status, the frequency of the tests was decreased to monthly blood work (liver function testing, hematology, biochemistry, and afp measurement) and medical consultation. Follow-up ct imaging was obtained 3 weeks after 90 Y radioembolization and then every 3 months.

Outcomes and Assessments
The primary outcome was safety and tolerability. Adverse events were recorded at each clinic visit, and patients were encouraged to report any change in health and symptoms. Secondary outcomes were tumour response, disease control rate, and overall survival. Tumour response was measured by imaging (ct or magnetic resonance imaging) according to the Response Evaluation Criteria in Solid Tumors, version 1.1 46 , before therapy start and after radioembolization; "disease control rate" consisted of stable disease and partial and complete responses. Overall survival was measured from the date of sorafenib treatment initiation to the date of death from any cause.

Statistical Analysis
In the statistical analyses, the Fisher exact test, chi-square test, and Mann-Whitney U-test were used, as appropriate. For continuous variables, means (averages) ± standard deviation were used for normally distributed data, and medians with interquartile range (iqr) were used for non-normally distributed data. Overall survival was assessed by Kaplan-Meier analysis. Any difference was considered significant when the p value was 0.05 or less. Statistical analyses were performed using the software

Patients and Exclusions
The study initially enrolled 40 patients, and 29 completed the treatment plan of combined sorafenib and 90 Y radioembolization ( Figure 2). In the initial cohort of 40 patients, mean age was 62 ± 11.6 years, and 37 patients were men (92.5%). The most common cause of liver disease was hepatitis C infection (32.5%, n = 13), and almost all patients had an ecog status of 0 or 1. The Barcelona Clinic Liver Cancer classification was B for 12 patients (30.0%) and C for 21 patients (52.5%); and 12 patients had previously received locoregional treatments (Table i).
The 11 patients who did not receive 90 Y radioembolization were ineligible because of large lung or bowel shunts, or variant vascular anatomy (5 patients); worsening ecog or patient status as assessed by the treating physician (4 patients); new portal vein thrombosis (1 patient) and new hepatic arterial thrombus (1 patient); and refusal to proceed with 90 Y treatment (1 patient). Of the excluded patients, 9 (81.8%) were classified Child-Pugh A, and 2 (18.2%) were classified Child-Pugh B.

Safety and Toxicity
Sorafenib discontinuation was required in 4 patients (13.8%), and sorafenib interruption in 4 patients, including 2 patients who required interruption before 90 Y treatment ( Table ii). The sorafenib dose had to be adjusted or reduced (to 200 mg twice daily) in 6 patients (20.7%, Table ii). If additional sorafenib dose reductions were required, patients took 200 mg every second day. The most common toxicities reported were fatigue (19.0%); alteration in liver function (7.9%, most commonly increased ascites); and diarrhea (6.3%). Most toxicities (63.3%, corresponding to 19 patients) were of grade 2 or less. Grade 3 toxicities (19.0% of all reported events) included abdominal pain (4.8%), alteration in liver function (4.8%), hand-foot syndrome (3.2%), diarrhea (1.6%), fatigue (1.6%), bleeding (1.6%), and alteration in coagulation (increased international normalized ratio, 1.6%). Most toxicities were reported before 90 Y radioembolization and were generally self-limiting or responded to dose adjustment. Grade 4 toxicities occurred in 2 patients (pulmonary embolism and progression of cirrhosis with liver failure). The 12 grade 3 toxicities were reported from 12 different patients, and 2 of those 12 patients were classified Child-Pugh B, the rest being classified Child-Pugh A. The grade 4 toxicities were reported from 2 other patients.
The 30-day mortality (after 90 Y radioembolization) was 3.4%, corresponding to 1 patient who died of liver failure, possibly related to treatment. When looking at the 6 patients classified Child-Pugh B who received combination therapy, grade 2 and 3 toxicities occurred in 3 and 2 patients respectively. Similar proportions were observed in the 23 Child-Pugh A patients who received combined therapy (grade 2 and 3 toxicities in 16 and 10 patients respectively). No mortality was observed in the 6 patients classified Child-Pugh B who received both radioembolization and sorafenib, and only 1 patient of those patients had received a score of 9, the remainder having been scored 7 or 8.  (Table iii). Compared with patients who were not responders (as defined by stable or progressive disease), those who experienced a disease response (complete or partial) after combined treatment trended toward having a smaller disease burden (total sum of lesion diameters) at baseline before therapy [median: 5.0 cm (iqr: 4.4-6.0 cm) vs. 9.3 cm (iqr: 5.2-11.9 cm); Figure 3], although the difference did not reach statistical significance (p = 0.1350). Most of the 90 Y treatments (82.8%) resulted in necrosis or reduced tumour enhancement.

Overall Survival
The median overall survival for the entire cohort of 40 patients was 11.9 months (iqr: 6.2-18.7 months). The

FIGURE 3
The burden of disease trended smaller in therapy responders than in non-responders, although the difference was nonsignificant. PR = partial response; CR = complete response; PD = progressive disease; SD = stable disease.

DISCUSSION
Although some retrospective data 48 have suggested that the combination of radioembolization and sorafenib therapy could be beneficial for advanced hcc, we believe that ours is the first prospective study to evaluate the combination of 90 Y radioembolization and sorafenib for unresectable hcc. Our preliminary results demonstrate the safety and tolerability of the two treatments in combination.
Almost three-quarters of the reported adverse events were grades 1 and 2, with 12 grade 3 events and only 2 grade 4 events reported. The reported symptoms were mostly similar to those described for the sharp trial (fatigue, hand-foot syndrome, rash and pruritus, and gastrointestinal symptoms such as diarrhea) and were thus possibly secondary to sorafenib. Worsening of liver function, observed as an increase in ascites or alterations in liver enzymes, was in contrast most likely related to the 90 Y radio embolization, because those events were comparable to the ones reported in earlier hcc and radioembolization studies 33,49 . Within the 30 days after 90 Y treatment, 1 patient died-a mortality rate similar to those reported in cohorts of hcc patients receiving radioembolization 49 . However, a fact worth noting is that the sorafenib dose had to be reduced for several patients (6 of 29) to 400 mg daily at maximum, and sorafenib treatment had to be interrupted at some point in the course of therapy for 8 patients. Those reductions and interruptions might suggest that the optimal dose would preferably be lower for patients undergoing this type of combined treatment, because half the patient cohort did not seem to be able to tolerate a full dose of sorafenib. That hypothesis also accords with results of a study by Kulik et al. 50 , which prospectively evaluated patients receiving 90 Y with and without sorafenib, and which also pointed toward a sorafenib dose reduction (400 mg daily), because dose reductions were required for all patients in the combined treatment arm.  creating a vasculature normalization window 38 . Since the introduction of angiogenic agents such as sorafenib for advanced hcc, there has been a theory that such agents could potentially "normalize" the tumour vasculature 54 , which is typically more disorganized. A sufficient duration of sorafenib treatment before the application of radioembolization (6-8 weeks was used in the present study) could theoretically allow for the development of the normalization window, thereby favouring a more effective radiation-induced tumour response. However, finding evidence for that hypothesis was not the aim of our study, and longitudinal studies evaluating the optimal normalization window would be needed to identify the best timing for administration of TheraSphere treatment.

CONCLUSIONS
Results of the present study show the safety and tolerability of combining 90 Y radioembolization with sorafenib for patient with unresectable hcc, although our results suggest that a lower dose of sorafenib might be administered when that treatment is given concurrently with radioembolization. The optimal timing of 90 Y treatment relative to sorafenib is still not well defined, and further investigations to objectively measure vascular normalization and to perform radioembolization at varying time intervals after sorafenib initiation might be warranted. Nonetheless, our data concerning tumour response and overall survival seem to show promising preliminary results. Our findings, in combination with larger prospective trials, should propose a new approach for hcc patients that could eventually be integrated into treatment algorithms.
For secondary outcomes, our study set out to assess tumour response, disease control rate, and overall survival. Treatment efficacy can be demonstrated by the tumour response assessment, which found that 82.8% of patients (n = 24) demonstrated radiologic evidence of necrosis, and by the disease control rate of 72.4% (n = 21). Patients experiencing partial or complete response tended to have a smaller burden of disease at baseline. Larger studies to identify variables associated with disease response would be warranted in an attempt to stratify patients, allowing for judicious selection of patients who might benefit most from this combination therapy.
Despite the nature of the present study as a pilot with a small number of patients, the combination of 90 Y and sorafenib was associated with a median survival duration of 12.4 months, which is comparable to the overall survival of 10.7 months reported in the sharp trial 14 . Moreover, subject selection in the sharp trial, which assessed the efficacy of sorafenib alone, favoured patients in better general condition (restricting enrollees to Child-Pugh A status, for example). The application of strict inclusion criteria was justified to avoid possible confounders with respect to causes of death, permitting the best possible assessment of the effect of sorafenib therapy alone on the tumours, and avoiding the effect on survival exerted by severe liver disease and cirrhosis. However, the indications for sorafenib do not exclude patients with more severe liver disease, and hence we opted not to exclude patients classified as Child-Pugh B and also to select patients based on functional status (ecog).
Although a direct comparison is difficult, our survival results also approach the neighbourhood of the survival duration data reported in 90 Y series, which range between 5 months and 17 months for Barcelona Clinic Liver Cancer classes B and C respectively 32,51,52 , thus representing most of the subjects included in our study. Our results, derived from a cohort of patients whose disease, at baseline, was more severe than that experienced by patients in the sharp trial, still showed comparable survival results. However, those survival data contrast with data from a study by Rana et al. 53 , who retrospectively reviewed 10 patients receiving sorafenib followed by radioembolization. In their study, which included 7 patients classified as Child-Pugh A and 3 classified as Child-Pugh B, the side effects with combined therapy were similar to those reported here-namely, anorexia, fatigue, gastrointestinal symptoms, and also the more severe side effects related to ascites and liver enzymes-and yet their reported overall survival duration was barely greater than 7 months, which is worse than the survival durations associated with both therapies alone. Those result could potentially be explained by the inclusion of patients at a worse disease stage or by the retrospective nature of the study, which can introduce confounding factors and make interpretation difficult. More insights about the specific side effects, toxicities, and tumour control data for each individual therapy might be provided by results from the sarah trial (NCT01482442 at http:// ClinicalTrials.gov), which compared the two treatments head-to-head, but which has yet to be published.
The rationale behind the combination of sorafenib and 90 Y radioembolization originates from the concept of