Arterial enhancement fraction in evaluating the therapeutic effect and survival for hepatocellular carcinoma patients treated with DEB-TACE

Hepatocellular carcinoma (HCC) is the third most common cause of cancer-related death worldwide [1]. Catheter-based locoregional treatment, also known as transarterial chemoembolization (TACE), is the most frequently implemented in patients with unresectable HCC across all disease stages [2]. The modified response evaluation criteria in solid tumors (mRECIST) was developed to assess response to TACE by measuring the shrinkage of viable tumor, seen as a decrease in contrast-enhancing areas at conventional contrast-enhanced imaging [3]. However, the size-based criteria do not take the quantity of enhancement into account and show impaired capability in some non-measurable HCC lesions [4, 5].

Computed tomography perfusion imaging (CTPI) provides quantitative information about the hemodynamics properties of tissue. The potential value of CTPI in response evaluation and prognostic prediction for HCC treated with TACE has been investigated by several studies [6‐9]. Among the various perfusion values that can be calculated with CTPI, hepatic perfusion index (HPI), the ratio of hepatic arterial perfusion to total perfusion, is regarded as an essential parameter for characterizing the hemodynamic features of HCC [10]. Liver CTPI typically involves scanning the liver at numerous (> 20) time points after IV contrast injection, thus requiring dedicated scanning protocols and a large radiation dose [11].

Arterial enhancement fraction (AEF) derived from routine triphasic liver CT examinations, introduced by Kim et al. [12], is an ideal surrogate biomarker for HPI and therefore addresses the issue of radiation dose. They defined AEF as the ratio of the absolute increment of attenuation in the arterial phase to that of the portal venous phase: AEF = [(HU_A - HU_U)/(HU_P - HU_U)] × 100%, where HU, A, P, and U stood for attenuation, arterial phase, portal phase, and unenhanced, respectively. Using the quantitative color mapping of AEF, they increased the sensitivity for HCC detection from 71.7 to 88.8% [12]. The strong correlation (r = 0.91, p <  0.001) between HPI and AEF was then observed in 10 rabbits with VX2 liver tumor by the same team [10]. Several authors have discussed the possibility of using AEF to assess the efficacy of chemotherapy, radiofrequency ablation, and radioembolization in liver metastases disease [13‐15]. Thus, we hypothesize that AEF would also be a feasible biomarker to evaluate the devascularization effect induced by TACE.

The current study aims to assess changes in AEF of viable tumor after drug-eluting bead (DEB) TACE. Furthermore, we seek to investigate the relationships between AEF and response and survival of HCC patients treated with DEB-TACE.

Demographics, underlying liver disease, and information about tumor-related details of the 158 patients are depicted in Table 1. Hepatitis B virus infection was the most common cause of HCC (n = 136, 86.1%), other causes included hepatitis C (n = 2), liver flukes infection (n = 3), cirrhosis caused by chronic Budd-Chiari syndrome (n = 1), and other unknown causes (n = 16). The degree of the macrovascular invasion incorporated Vp1 or Vp2 portal vein tumor thrombosis (PVTT) according to the Japanese Society of Hepatology PVTT classification [20] (n = 56) and hepatic vein tumor thrombosis (n = 3).

During the evolution of dysplastic nodules to HCC, there was a pattern of hemodynamic change, which involved increasing arterial blood supply due to tumor-related angiogenesis [21]. CTPI enabled the quantification of perfusion characteristics in tumor tissue and has proved a potential technique for assessing the efficacy of various HCC treatments [22‐24]. However, the high radiation exposure was one of the most problematic issues limiting the application of this technique, particularly considering that cancer patients may need to undergo repetitive imaging examinations to monitor treatment response [6]. Hepatic AEF, derived from routine triphasic CT scans, was an ideal biomarker that allowed indirect estimation of hepatic arterial perfusion to total perfusion without raising the extra radiation concern. Several studies have demonstrated the application value of AEF on chemotherapy, radiofrequency ablation, and radioembolization in liver metastases disease [13‐15].

Our study assessed the changes in AEF of HCC after DEB-TACE and investigated the relationships between AEF and response and survival outcome. The results showed that, before embolization, only AEF of HCC in patients with Child class B was significantly higher than in those with Child class A, while no correlations were observed between AEF and serum marker or tumor characteristics such as diameter. Kaufmann et al. [25] used a CT-based volume perfusion technique to characterize HCC lesions. The result suggested that HPI (a perfusion parameter similar to AEF) was not correlated with lesion size, which was supported by our finding (r = 0.05, p = 0.551). Kang et al. [26] reported that the AEF of liver parenchyma in Child class A group resembled that in Child class B groups (23.7% ± 7.6% vs. 32.2 ± 10.9%, p = 0.16). Though we got a similar result to Kang et al.’s, the difference in AEF between two groups was much less evident in our investigation (19.1% ± 7.9% vs. 20.0% ± 5.4%, p = 0.498). It may be attributed to the fact that ROI was delineated based on each liver segment, significantly larger than the ROI area in this study. More importantly, the present study failed to find the association between AEF and treatment response. In other words, no prediction of response was possible before DEB-TACE based on AEF. Nevertheless, Mao et al. [27] succeeded in predicting treatment response by using texture features of HCC on the AEF color map.

The relatively large size of our sample allowed us to divide the patients into different treatment response groups to compare the changes in AEF among them. In the CR group, lesions on their post-treatment color mappings displayed as signal loss areas so that neither AEF_post nor decrease ratio of AEF could be determined. In both PR and SD groups, the AEF of the residual tumor was significantly lower than that of the same lesion before embolization (Table 2). However, the decrease ratio of AEF in patients with SD response was significantly lower than in those with PR response (Fig. 3). In the PD group, the AEF of the residual tumor was even numerically higher than that of the original lesion before treatment (Table 2). Correlation analysis also showed that the decrease ratio negatively correlated with treatment response (ρ = − 0.47, p <  0.001). The trend in AEF across the response groups was similar to previous CTPI research conducted by Chen et al. [24]. They observed, after TACE treatment, the hepatic arterial fraction (HAF, a perfusion parameter used in GE Medical perfusion software, with the exact definition as HPI but calculated with deconvolution model) of HCC lesion significantly decreased in the PR group (63.7% vs. 38.2%, p = 0.030), and increased in PD group (45.9% vs. 69.6%, p = 0.012), respectively. In the SD group, HAF of residual tumor was numerically lower than HAF before treatment, though the difference did not reach statistical significance (42.3% vs. 33.8%, p = 0.248). The possible explanation was that incomplete embolization would cause local hypoxia in the tumor microenvironment, leading to the upregulation of multiple angiogenesis factors and the development of new arterial vessels that sustained tumor relapse. Further histological research was needed to prove such speculation.

Although there was no way to predict treatment response using baseline AEF, we managed to determine cutoff values for AEF_pre, AEF_post, and decrease ratio to identify patients with favorable and unfavorable OS (Fig. 4). The Cox regression analyses suggest that relatively high AEF_pre and AEF_post result in discouraging survival outcomes. As for the impact of the decrease ratio on the OS, we find that patients would be more likely to gain a significant survival benefit if there is any reduction in tumor AEF after TACE treatment. Surprisingly, the risk remains reduced in patients who obtained an increased AEF with a ratio not exceeding 0.102. However, the impact of the decrease ratio on OS was no longer significant after being adjusted by confounding factors (Table 4).

Other than all of the inherent defects in this research design, our work had some other limitations. Though we speculated that the correlation between AEF and response for DEB-TACE was the consequence of tumor-related angiogenesis, further histological research was needed to prove such speculation. Patients who selected conventional TACE as initial treatment were not included in this study due to the artifact induced by Lipiodol. An MR-based AEF analysis may solve the problem.

This study revealed that no prediction of response was possible before DEB-TACE based on AEF. After DEB-TACE, the AEF of residual viable tumor and decrease ratio of AEF could be the easily accessible parameters for monitoring the response of HCC to DEB-TACE. In addition, AEF is associated with OS in HCC patients treated with DEB-TACE, which could be a candidate prognostic factor in future studies.