Yttrium-90 Radioembolization for Hepatocellular Carcinoma with Portal Vein Invasion: Validation of the Milan Prognostic Score

Clinical question
Does the Milan portal vein tumor thrombosis (PVTT) score represent a useful tool in selecting patients with hepatocellular carcinoma (HCC) with PVTT who may benefit from radioembolization?

Take away point
This study supports the use of the Milan PVTT score as a clinical tool to identify subgroups of patients with HCC and concurrent PVTT who may benefit most from transarterial radioembolization (TARE), demonstrating an overall survival (OS) of greater than 2 years in the properly selected patient.

Reference
Yttrium-90 Radioembolization for Hepatocellular Carcinoma with Portal Vein Invasion: Validation of the Milan Prognostic Score. Bargellini, Irene et al. Journal of Vascular and Interventional Radiology, Volume 31, Issue 12, 2028 - 2032

Click here for abstract

Study design
Retrospective single-center cohort study including 70 patients with advanced HCC and PVTT.

Funding Source
Self-funded or unfunded

Setting
Academic hospital, Pisa University Hospital, Pisa, Italy.

Figure

Figure: OS according to Milan PVTT prognostic scores: group A, good prognosis; group B, intermediate prognosis; group C, dismal prognosis.

Summary

Although advanced-stage HCC is traditionally treated by systemic therapies, there is ongoing debate about the role of locoregional therapies for these patients, in particular those with PVTT. To date, prospective randomized trials have not demonstrated an advantage of TARE versus sorafenib systemic therapy in intermediate to advanced stage HCC. The Milan PVTT score has been recently proposed as a stratification model to better predict outcomes in those with advanced HCC and PVTT undergoing yttrium-90 (Y90) TARE; three parameters including bilirubin levels, extent of PVTT extension, and tumor burden delineate 3 subgroups with good, intermediate, and poor prognosis. As previously described, using this model to select Y90 TARE candidates with good liver function and limited PVTT showed a median OS time as long as 30 months. This study was undertaken using a slightly modified Milan PVTT score, in an effort to further validate the stratification model.

Patients with advanced HCC and PVTT were selected based on the following criteria; presence of HCC with PVTT diagnosed according to the European Association for the Study of the Liver criteria, PVTT extending to the main portal trunk but not exceeding the level of the hepatic hilum, absence of extrahepatic metastases, Child-Pugh class/score no worse than B/7, Easter Cooperative Oncology Group (ECOG) performance status 0/1, and age > 18 years. A total of 70 patients were included, some of which had previous locoregion, surgical, or systemic HCC treatment.

PVTT was defined as visualized filling defects in the portal venous phase with enhancement during the arterial phase. PVTT was graded as segmentary (PV1, score of 0), second-order venous branch (PV2, score of 2), first-order branch (PV3, score of 3), and main portal trunk (PV4, score of 4). Patients were stratified using a Milan PVTT score modified to include PV4. Additionally, PVTT score was summed with bilirubin level (≤ 1.2 mg/dL [score of 0], > 1.2 mg/dL [score of 2]), and tumor burden (≤ 50% [score of 0], >50% [score of 3]) to grade patients into a good prognosis (total score of 0), intermediate prognosis (total score of 2 or 3), and poor prognosis (total score > 3). Fifteen patients (21.4%) were in the good prognostic group, 33 (47.1%) in the intermediate prognostic group, and 22 (31.4%) in the poor prognostic group.

TARE was preceded by a standard simulation procedure using technetium-99m–labeled macroaggregated albumin to map arterial feeding vessels and characterize shunting to nontarget organs. Both Y90 resin (SIR-Sphere, used in 64.3% of patients) and glass (TheraSphere, used in 35.7% of patients) were administered by 2 interventional radiologists. Followup included triphasic CT and/or magnetic resonance imaging with routine clinical and laboratory follow-up every 4-6 weeks, followed by every 3 months thereafter. Tumor response was based on a modified Response Evaluation Criteria in Solid Tumors, defined as complete response, progressive disease (PD) in unequivocal progression, and non-PD in all other patients. Data was analyzed using the mean and standard deviation compared with χ2 or Fisher exact tests for categorical data and Student t test for paired data. OS and progression-free survival were calculated using the Kaplan–Meier method, and survival curves were compared by log-rank test.

Median OS was significantly (P = 0.0012) different according to modified Milan PVTT score: 24.6 months in the good prognosis group, 13 months in the intermediate prognosis group (hazard ratio, 3.2; 95% CI, 1.2–9.7; P = 0.016), and 5.9 months in the poor prognosis group (hazard ratio, 4.1; 95% CI, 1.4–13.4; P = 0.0096; See Fig). Complete response was found in 3 cases (4.5%), partial response in 33 cases (50%), stable disease in 21 cases (31.8%), and PD in 9 cases (13.6%). The early objective tumor response rate and median progression-free survival were not significantly different across prognostic groups. In the case of PD or residual viable tumor, patients were treated according to disease presentation after multidisciplinary tumor board discussion. Treatment included additional TARE, chemoembolization, percutaneous ethanol injection, systemic therapy, or second-line experimental systemic therapies.

Overall, the authors think their results externally validate the previously created Milan PVTT score. The authors directly compare their results to multiple similar studies which reported similar OS stratification. The score allows easy grouping of patients with advanced-stage HCC with PVTT who could most benefit from radioembolization.

Commentary

The authors identify some key limitations inherent to their retrospective study, including possible selection bias in patients selected for radioembolization after multidisciplinary discussion, heterogeneous patient clinical history and previous treatment, the two types of spheres used for TARE, and variable follow-up treatment used to treat residual tumor or control tumor progression. I agree that the heterogeneity of the patients and their follow-up treatment planning certainly limits the scope of the study.

Ultimately, this study serves to validate and slightly modify the Milan PVTT model, but has little clinical significance in and of itself. The authors astutely identify that this model could represent the basis for future studies comparing locoregional approaches such as TARE versus systemic therapies, as well as evaluating future therapeutic combinations. This simple model can serve as a foundation for future integration of additional tumor data, such as pathology, genomics, and derived radiomics, contributing to increasingly sophisticated means of stratifying patients for various treatments.

Post Author
Gregory Rufener, MD
Radiology Resident (PGY-3)
Department of Radiology
Oregon Health and Science University, Portland, Oregon
Twitter: @gregoryrufener

Edited and formatted by @NingchengLi