Treatment of Hepatocellular Carcinoma-associated Portal Vein Tumor Thrombus Using Combination of Portal Vein Stent Insertion and Endovascular Lodine-125 Seed-strip Implantation Followed by Transcatheter Arterial Chemoembolization-Sorafenib

Background To evaluate the ecacy of portal vein stent (PVS) insertion and endovascular iodine-125 ( 125 I) seed-strip implantation followed by transcatheter arterial chemoembolization (TACE)-sorafenib (PVS- 125 I-TACE-S) in patients with hepatocellular carcinoma (HCC)-associated portal vein tumor thrombus (PVTT). Methods This retrospective analysis was performed on 53 consecutive HCC patients with PVTT between May 2014 and July 2018. Twenty-eight patients were treated via PVS- 125 I-TACE-S, and 25 patients were treated by TACE-S. Hepatic function, disease control, and overall survival (OS) in both groups were analyzed. The Albumin-Bilirubin (ALBI) score approach was used as a means of evaluating liver function, and a Cox regression analysis was utilized to determine which factors were linked to treatment outcomes.

choice for individuals with advanced HCC, irrespective of PVTT status. The Sorafenib Hepatocellular Carcinoma Assessment Randomized Protocol (SHARP trial), on which BCLC recommendations were based, however, suggests that vascular invasion is only present in 38.4% of HCC patients, and as such sorafenib alone may not necessarily confer improved survival to HCC patients with PVTT [6,7]. In addition, studies of Asian populations with vascular invasion or extrahepatic metastasis have suggested that sorafenib treatment is associated with poor overall survival (OS; median: 6.5 months) [6,8]. Indeed, consistent with this the Asia-Paci c guidelines recommend that HCC patients with PVTT undergo a multipronged antitumor treatment course that can include surgery, sorafenib, and radiotherapy or transcatheter arterial chemoembolization (TACE) as potential interventions [9]. There has been some research regarding the use of TACE-sorafenib (TACE-S) owing to observations that it exhibits therapeutic synergy when employed to treat HCC patients with PVTT provided there is good patency for the main portal vein or su cient collateral circulation [6,10,11]. This combination is associated with some increase in patient survival, but overall the e cacy of this approach is limited, and a superior combination strategy is thus needed [12].
The use of a portal vein stent (PVS) to reduce portal vein occlusion has been explored in the context of PVTT, and its combination with iodine-125 ( 125 I) seed brachytherapy offers a potentially optimal means of improving PVTT management in HCC patients [13]. There are previous reports showing that this combination could decrease liver toxicity caused by TACE, and the synergic effect or abscopal effect offers better local control and OS [14,15]. Therefore, PVS insertion and endovascular 125 I seed-strip implantation followed by TACE-S has a strong scienti c rationale [16]. In the present study, a retrospective analysis was conducted to assess the e cacy of this approach and the factors associated with patient outcomes.

Methods
This study received approval from the ethics committee of our institution, and all provided written informed consent before participating.
The institutional database used in this study was compiled using data from HCC patients with PVTT who underwent endovascular 125 I seed-strip implantation followed by TACE-S (PVS-125 I-TACE-S) or TACE-S from May 2014 to July 2018. For all patients, attending physicians had recommended TACE, sorafenib, TACE-S, or PVS-125 I-TACE-S treatment, with nal treatment approaches having been selected by the individual patients. Study inclusion criteria were as follows: 1) diagnosis with HCC and either type II or type III PVTT; 2) no previous local treatment for PVTT lesions; 3) An Eastern Cooperative Oncology Group (ECOG) performance of 0-2; 4) Child-Pugh Class A or B. Study exclusion criteria were as follows: 1) complete portal vein occlusion with a lack of collateral circulation; 2) bleeding of the esophagus or gastric fundus; 3) intractable coagulation disorders; 4) macroscopic hepatic vein tumor thrombi or extrahepatic tumor metastasis; 5) a lack of baseline imaging results.
A total of 64 consecutive HCC patients with PVTT were identi ed, including 7 patients that underwent simple TACE, and 1 patient treated with only sorafenib. In addition, preoperative enhanced CT/MRI scan results were not available for 2 patients, and 1 patient was lost to follow up. These patients were excluded from the study, leaving a nal study population of 53 patients . A total of 28 of these patients   were treated via PVS-125 I-TACE-S (group A), while the remaining 25 patients were treated via TACE-S alone  (group B).
In all patients, HCC diagnosis was con rmed either via histology or by a combination of two imaging approaches and elevated α-fetoprotein, as per the American Association guidelines [17]. Magnetic resonance imaging (MRI) or computed tomography CT) was used to diagnose PVTT based on the presence of a low attenuation intraluminal mass that partially or fully occluded the portal vein or lled a portal vein defect [18]. Patients were strati ed into Type I -IV PVTT according to Cheng's PVTT classi cation system based on the following criteria: Type I, tumor thrombi locating at or above portal vein segmental branches (second-order branches); Type II, tumor thrombi extending so as to involve the right or left portal vein ( rst-order branches); Type III, thrombi involving the main trunk of the portal vein; Type IV, invasion of the superior mesenteric vein or inferior vena cava [4].

PVS combined with endovascular 125 I seed-strip implantation
The required number (N) of 125 I seeds was determined based upon the length (L) of the obstructed portal vein segment in mm per the formula N = L/4.5 + 2 (Shanghai GMS Pharmaceutical Co. Ltd) [13]. The 125 I seeds used in this study were 0.8 mm in diameter, and 4.5 ± 0.5 mm long, with 25.9 MBq of radioactivity and a 59.4-day half-life, primarily emitting 27.4 and 31.4 keV X-rays and 35.5 keV -rays. Given the local tissue half-value thickness of 17 mm, these seeds were associated with an initial dose rate of 7 cGy/h. Prior to implantation, these seeds were loaded in a linear arrangement in a 4 Fr exible stiffening cannula (Boston Scienti c Co., Ltd). With ultrasonic guidance, a Neff Percutaneous Access Set (Cook, Inc., IN, USA) was then used to puncture the patent second-order portal vein branch, after which both a vascular stent (diameter: 12-14 mm; length: 60-100 mm; Bard Peripheral Vascular Inc., AZ, USA) and 125 I seed-strip were implanted in succession. After this, a 3-3 spring coil (Cook, Inc., Bloomington, Indiana) was used for blocking the intrahepatic puncture, and subcutaneous low-molecular-weight heparin (4100 IU; Hebei Changshan Biochemical Pharmaceutical Co, Ltd) was administered twice daily for a 5 day period, after which warfarin (Shanghai Sine Pharmaceutical Laboratories Co., Ltd) was administered orally in order to achieve the international normalized ratio of 2.0-2.5.

TACE procedure
In group A patients, TACE was conducted 3 -7 days after PVS and seed implantation, whereas in group B patients this procedure was conducted directly. A 5-Fr hepatic-curve catheter (Terumo Corporation, Japanese) was placed into the celiac artery and then hepatic arterial angiography and indirect portography were performed. Next, a 2.7 Fr microcatheter (Progreat™, Terumo, Tokyo, Japan) was placed into tumor-feeding arteries, and 5-20 mL lipiodol (Beijing Wh-Medical Apparatus and Instruments Co.
Ltd) mixed with 50-75 mg/m 2 doxorubicin hydrochloride (Shanghai Hisun P zer Pharmaceutical Co. Ltd) were injected into these arteries. If the entire 20 mL volume was administered without substantially impairing blood ow in these vessels, polyvinyl alcohol (PVA) particles (Hangzhou ALICON Pharmaceutical Science and Technology Co. Ltd) were used for vessel embolization, being administered until only a limited slow ow was evident. For those with arterioportal shunts, initial embolization using 350-1000 μm PVA particles was conducted prior to lipiodol/doxorubicin infusion in order to ensure shunt occlusion. TACE treatment was guided by imaging results.

Sorafenib
Three to seven days after TACE treatment, when liver function had stabilized, patients were administered 400 mg sorafenib (Nexavar®, Bayer HealthCare, Leverkusen, Germany) twice per day. In many patients suffering from grade 1 or 2 adverse events (AEs), or with serum bilirubin > 34.2 μmol/L, sorafenib doses were reduced. In patients suffering grade 3 AEs or hyperbilirubinemia (serum bilirubin > 51.3 μmol/L) events, sorafenib treatment was temporarily halted and was only resumed if AE grades and serum bilirubin levels fell back within the acceptable treatment ranges. Patients were continuously administered sorafenib tablets until it was no longer possible to do so, even when disease progression was noted, or until death.
Follow-up and treatment evaluation At 1 and 3 months post-operatively, patient follow-up was conducted, and it was then performed every three months thereafter. During follow-up, patients underwent physical examination, laboratory analyses, as well as contrast-enhanced abdominal MRI or CT. The Common Terminology Criteria for Adverse Events (CTCAE) v4.0 assessment was used to gauge any complications associated with treatment [8]. Albumin-Bilirubin (ALBI) scores were used as an objective means of gauging liver function, with scores being calculated solely based upon albumin and bilirubin levels [ALBI score=log10 bilirubin (μmol/L) ×0.66) + (albumin (g/L) ×−0.0852)] [17]. The modi ed Response Evaluation Criteria in Solid Tumors (mRECIST) criteria for HCC were used to gauge tumor responses, with possible responses including complete response (CR), partial response (PR), stable disease (SD), or progressive disease (PD) (Fig. 1). A disease control rate (DCR) after 6 months of treatment was then calculated based on the formula: (CR+PR+SD)/total cases×100% [19]. OS was the study endpoint and was de ned as the period of time between treatment and either death or last follow-up.
Statistical analyses SPSS 21.0 (SPSS Inc., Chicago, IL, USA) was utilize for all testing. Quantitative results are given as means ± standard deviation, and t-tests was used to compare these values. Qualitative results are given as a number (%), and Pearson Chi-Squared tests or Fisher's exact tests were employed for comparing these results as appropriate. Paired sample t-tests were used for comparison of ALBI scores and pre-and post-operative portal venous pressure in group A patients. Survival was analyzed via the Kaplan-Meier approach and through log-rank tests. P<0.05 was the signi cance threshold. Cox univariate regression analyses were used to assess the relationship between speci c factors and treatment outcomes. Multivariate logistic regression analyses were conducted based on the variables in the univariate analysis (P<0.10).

Baseline information
Baseline characteristics of the 53 patients in this study are shown in Table 1, with no signi cant clinicallyrelevant differences between groups. For group A patients, on average 18.2 ± 1.7 (range: [16][17][18][19][20] 125 I seeds were implanted, with mean radioactivity of 12.6 mCi ± 1.2 (range: 11.2-14.0 mCi). Estimated radiation doses were roughly 40 -50 Gy, with a computerized treatment planning program (FTT Technology Ltd. Co, Beijing, China) being used to determine speci c dose reference points.

Complications and clinical outcomes
There were no observed instances of abdominal hemorrhage, biloma, liver abscess, stent migration, or puncture site bleeding following PVS and endovascular 125 I seed-strip implantation in any patients. In these patients, ALBI scores did alter following treatment, changing from -2.57 ± 0.42 to -1.92 ± 0.53 (P<0.001). TACE-S-related AEs did not differ signi cantly between groups (P>0.05) ( Table 2).
Univariate Cox model analyses suggested that tumor size and number were both variables potentially associated with treatment outcomes, and as such, these were incorporated into a multivariate model. This analysis, in turn, determined that tumor size > 10 cm was an independent predictor of poor prognosis (P=0.002), whereas PVS-125 I-TACE-S was a predictor of a favorable patient prognosis (P=0.040) ( Table 4).

Discussion
HCC often readily undergoes portal vein invasion, leading to the development of PVTT, which can partially or fully obstruct the vein, resulting in increased venous pressure, decreased perfusion, and ultimately impaired liver function. The resultant elevation in pressure within the portal vein is also often associated with esophageal or stomach fundus bleeding. For these reasons, HCC patients with PVTT often have a poor prognosis [20]. While liver resection can cure some patients with type I-III HCC [21,22]. However, these patients typically exhibit poor long-term survival following surgery as HCC commonly recurs, particularly in the case of early recurrence within 1 year post-surgery [23].
One approach to improving HCC/PVTT patient OS is radiofrequency ablation, but because PVTT is near the bile duct, this approach has a signi cant potential to cause injury to either this structure or to the portal vein itself [24]. Also, transarterial radioembolization (TARE) has been used more frequently in patients with HCC and PVTT due to the decreased risk of hepatic ischemia and infarction, but it remains unavailable in commercial contexts in Japan and mainland China [25]. Bene ts of stereotactic body radiotherapy (SBRT) include the fact that tumors can receive higher direct radiation doses while organs are protected from substantial radiation. However, lesion size is the main limiting factor for this approach, with local control rates ranging from 91% (<5 cm tumors) to 74% ( 5 cm tumors) [26]. As such, many therapeutic strategies have been employed to treat patients suffering from HCC and PVTT, including TACE, sorafenib, and 3-dimensional conformal radiotherapy or combinations thereof. Even so, at present none of these approaches have been able to reliably and rapidly achieve desired treatment outcomes [27].
In recent years, brachytherapy has been widely utilized for tumor treatment. Due to the unique radiobiological characteristics of 125 I seed, it effectively killed tumor cells without obvious damage to surrounding normal tissue [24]. In the present study, a signi cant survival bene t in patients in group A relative to those in group B (342.0 vs 231.0 days). This may be because patients treated with the combination of PVS and endovascular 125 I seed-strip implantation prior to TACE-S achieved better liver function, suggesting that restoring portal vein through and reduction of tumor thrombus burden is associated with improved OS [12,13]. With respect to safety, the most frequent AEs following the combination of PVS and 125 I seed-strip implantation was a transient decrease in liver functionality, likely as a result of injury to the bile duct upon puncture of the portal vein [28]. TACE-S-associated adverse events for the two groups did not signi cantly differ. Grade 1/2 AEs were mild and alleviated through symptomatic treatment. Grade 3 and 4 AEs were managed in a conservative manner, temporarily reducing or ceasing sorafenib dosing as appropriate. Together these ndings suggest that PVS combined with endovascular implantation 125 I seed-strip followed by TACE-S is a well-tolerated intervention strategy.
Preoperative ALBI scores were comparable between groups (P=0.724), nor did these scores differ 1month postoperatively (P=0.666). However, at 3 (P=0.007) and 6 (P=0.044) months postoperatively, these scores were signi cantly different. PVS is well known to provide immediate restoration of blood ow to an obstructed portal vein, improving the hepatic blood supply. In addition, as 125 I seed-strip were used for sustained intravascular brachytherapy, they may be a suitable means of counteracting neointimal hyperplasia and improving stent patency duration [28]. Together, these therapeutic interventions may thus be able to better ensure prolonged liver functionality following repeated TACE and long-term oral sorafenib intake. In contrast, group B patients only treated by TACE-S, where the embolization is linked to a higher risk of ischemic necrosis of healthy tissue, and chemotherapy can additionally aggravate remaining liver function [29]. And sorafenib itself also has the potential to reduce portal blood ow, elevating the risk of liver failure [30].
Also, a better disease control in group A than in group B (71.4% vs 44.0%) after 6 months of treatment. TACE is reported to block ow through arteries that supply blood to the tumor, thus allowing for control of tumor growth and PVTT progression as the hepatic artery is the primary source of blood for tumor cells and thrombi [17]. As this region has blood supplied by both the artery and portal vein, however, and due to the potential for collateral arterial development, complete local necrosis is often di cult to achieve [31,32]. TACE is also linked to increased levels of vascular endothelial growth factor (VEGF), resulting in a higher risk of local recurrence [31]. While sorafenib combination treatment can reduce TACEassociated risks and improve outcomes, there are reports suggesting that sorafenib is less effective in those with Vp3/4 PVTT (< 10% response rate) [11,33]. In contrast, 125 I seed implantation in the portal vein can allow for effective PVTT control owing to sustained low-dose x-and γ-ray release throughout the tumor area, damaging tumor cell DNA and disrupting proliferation [24,34].
In a subgroup analysis, a signi cant difference in group A patient OS only for those with type III PVTT (344.0 vs 226.0 days). This may be due to the fact that near-total portal vein blockage in type III PVTT patients is linked to a rapid decrease in liver function [35]. And the liver function can be aggravated by repeated TACE and long-term sorafenib, as described above. This, in turn, requires modi cations to or discontinuations of TACE-S treatment regimens, thus potentially constraining their therapeutic value [36,37]. Patients with PVTT also often died of acute variceal bleeding rather than due to direct damage from the liver tumor itself [32]. The PVS/ 125 I seed implantation strategy can improve portal vein obstruction, thereby reducing the risk of esophageal and gastric bleeding, thus reducing the risk of this indirect cause of death [38,39]. In addition, 125 I seed implantation can improve local tumor control and delay PVTT progression, thus likely reducing the risk of local and distant metastasis [15].
No signi cant differences in OS for type II PVTT patients in these two treatment groups (313.0 vs 321.0 days), despite better liver function scores for group A patients. This may be because those in group A had a larger tumor burden, that tumor size > 10 cm was an independent predictor of poor prognosis (P=0.002). This further suggests that tumor burden is a key prognostic indicator in HCC patients with PVTT and must be considered during treatment planning [40].
There are certain limitations to this study. For one, sample size was small, and while these results were statistically signi cant, a future larger longitudinal study would be of value for con rming these ndings.
This was also a retrospective analysis, and as such future randomized controlled trials will be essential in order to validate these ndings. Lastly, a skin surface dosimeter was not employed after 125 I seed-strip implantation, making it impossible to determine the exact radiation doses administered to patients in this study.

Conclusions
The combination of PVS and endovascular 125 I seed-strip implantation followed by TACE-S represents a safe and effective treatment for HCC patients suffering from type III PVTT, potentially being more effective than TACE-S alone.

Consent for publication
Not applicable.

Availability of data and materials
Research data are stored in an institutional repository and will be shared upon request to the corresponding author.

Competing interests
The authors declare that they have no competing interests.   TACE-S, transcatheter arterial chemoembolization-sorafenib. P<0.05 was the signi cance threshold.