Radiation-induced liver disease after stereotactic body radiotherapy for small hepatocellular carcinoma: clinical and dose-volumetric parameters

Hepatocellular carcinoma (HCC) is one of the most common malignancies, ranking sixth worldwide[1]. Surgical resection and liver transplantation are the primary treatment modalities for HCC[2‐4]. However, strict criteria limit the pool of eligible patients for both approaches. Radiofrequency ablation (RFA) can also be used with curative intent for small HCC which is unsuitable for surgery[5, 6]. However, the proximity of the tumor to the main blood vessels, gall bladder, diaphragm, and liver surface presents major restrictions to use of RFA. Consequently, overall, only 30–40% of patients may derive benefit from the various radical curative treatments due to the numerous clinical conditions[7]. Therefore, in these clinical settings, alternative local therapeutic modalities are urgently needed.

With the development of three-dimensional conformal radiotherapy (3D-CRT), image-guided treatment and the resulting accumulation of knowledge on partial-volume liver tolerance, high-dose radiation could be delivered to focal liver volumes, thereby allowing radiotherapy as an alternative option for small intrahepatic tumors[8]. Several studies demonstrated that stereotactic body radiotherapy (SBRT) with 24–60 Gy administered in three to six fractions achieved high local control and overall survival rates of 65–100% and 48–82% at 1 year, respectively[9‐13]. While many studies have demonstrated that variable factors are associated with radiation-induced liver disease (RILD) after conventional radiotherapy, only a few have examined RILD after SBRT and their results focused on clinical outcome and local control[8, 12, 14‐17]. Recently, one study was aimed at identifying the parameters to predict hepatic toxicity and the deterioration of hepatic function[18]. However, additional studies are warranted to confirm the previous results.

Thus, the aim of the present study is to analyze the clinical and dose–volumetric parameters that predict the risk of RILD in patients with small, unresectable HCC treated with SBRT.

The present study evaluated a relatively large number of patients with HCC who received high-dose irradiation (≥ 10 Gy per fraction). Upon multivariate analysis, only the Child-Pugh class was of statistical significance in predicting the risk of RILD development. By contrast, the optimal dose–volumetric parameter predictive of RILD could not be determined. The baseline liver condition is thought to be the most important factor associated with the risk of RILD after high-dose radiotherapy, even if the irradiated volume is small.

Our results showed that patients with Child-Pugh B cirrhosis more frequently developed RILD than did patients with Child-Pugh A liver function. Similar results were obtained in some previous studies of conventional radiotherapy for primary liver tumors, which concluded that baseline liver function was an important factor in predicting the occurrence of RILD[15, 20]. The suggested mechanism was that a severely cirrhotic liver is less tolerable to irradiation because cirrhosis prevents the repair of radiation injury as well as the cellular proliferation, which is required to compensate for the loss of hepatic function. An association between the risk of RILD and liver function status after SBRT for HCC has been found in a limited numbers of studies. Mendez Romero et al. conducted the prospective trial of SBRT in the treatment of primary liver tumors and prescribed different dose-fraction schedules according to the tumor size and presence of cirrhosis. Acute toxicities ≥ grade 3 were observed in four patients, and one patient with HCC and Child-Pugh B liver function developed liver failure together with an infection and died. The authors therefore concluded that extreme caution is required for patients with Child-Pugh B cirrhosis due to high risk of toxicity[11]. In the series of Cardenes et al., the prescribed dose was increased to 48 Gy (16 Gy/fraction) in patients with Child-Pugh A cirrhosis without dose-limiting toxicity, however, two patients with Child-Pugh B liver function developed CTCAE grade 3 hepatic toxicities at the 42 Gy (14 Gy/fraction) level[9]. The results of these studies further support liver function as one of the most important factors in RILD, even though the irradiated volume treated with SBRT was much smaller than that targeted by 3D-CRT.

Several dose–volumetric parameters associated with RILD after 3D-CRT has been reported in the literature; however, reports evaluating acute and late hepatic toxicities after SBRT for HCC are scarce. Rusthoven et al. reported hepatotoxicity after 36–60 Gy, administered in three fractions, for the treatment of liver metastases, none of the 47 patients experienced late hepatotoxicity[12]. Lee et al. also evaluated RILD after SBRT in 68 patients with metastatic liver tumors who received a median dose of 41.8 Gy in six fractions and reported no occurrence of RILD or other grade 3–5 liver toxicity[14]. Notwithstanding these results of metastatic liver tumors, dose–volumetric parameters to predict RILD in patients with HCC who underwent SBRT were also reported. Son et al. reported that 12 patients developed grade 2 or higher hepatic toxicity after SBRT for small HCCs in 36 patients. This study focused not only on RILD but on the progression of Child-Pugh class as an endpoint for toxicity. The only significant parameter associated with the progression of Child-Pugh class (4 patients, 11%), as determined by multivariate analysis, was the total liver volume receiving a dose < 18 Gy. The authors therefore recommended that the total liver volume receiving < 18 Gy should be > 800 cm³, to reduce the risk of a deterioration of hepatic function[18]. Table 5 summarizes the toxicities reported in several studies of SBRT for primary and metastatic liver tumors.

The liver function of patients with HCC is usually already compromised prior to tumor development due to the pre-existing cirrhosis after chronic liver disease, whereas this is generally not the case in patients with metastatic liver tumors. Accordingly, the QUANTEC defines a non-classic RILD and recommends that this endpoint could be appropriate in HCC patients who have poor liver function including HBV infection. They also suggest that CTCAE is most useful for scoring non-classic RILD[19]. This is the reason why we adopt non-classic RILD as the endpoint of the present study. However, we cannot neglect classic RILD which occur occasionally in patients with HCC after radiotherapy and modify the endpoint as described previously. Up to now, the definition of RILD in patients with HCC was somewhat different among the studies. Moreover, a confounder, such as the progression of cirrhosis regardless of any treatment, is an important hurdle to define RILD especially in populations with pre-existing liver dysfunction[19]. For a definite answer to RILD after SBRT, we have to overcome many obstacles to the definition of toxicity and exact meaning of laboratory findings in the follow-up examination.

In the present study, a significant dose–volumetric parameter, including mean liver dose, associated with RILD could not be identified on multivariate analysis. This is partly because SBRT is a safe modality even in patients with underlying liver disease. The median GTV (8.6 cm³, range 0.6–125.3 cm³) was smaller than that reported in the study of Son et al. (18.3 cm³, range 3.0–81.3 cm³) and the incidence of RILD ≥ grade 2 was also lower. Therefore, the incidence of RILD was not increased significantly, even though the number of patients treated with SBRT was increased. This is one of the reasons for the low statistical power of the present study. However, this is not sufficient to explain the assumption. Another possibility is that RILD may be induced by unknown causes including the production of cytokines, regardless of the irradiated normal liver volume. Hence if the mentioned assumption is true, the RILD can occur randomly after SBRT for small HCC. A dose–volumetric parameter associated with the risk of RILD could not be determined, even based on a definition of RILD as mild toxicity (grade 2). There are also no studies suggesting a normal tissue complication probability (NTCP) model to predict the risk of RILD after SBRT at doses ≥ 10 Gy per fraction. In light of the small number of cases of RILD, further studies in a larger series will be needed to construct an appropriate NTCP model.

In conclusion, the Child-Pugh B cirrhosis was found to have a significantly greater susceptibility to the development of grade 2 or worse RILD after SBRT in patients with small, unresectable HCC. Additional efforts aimed at testing other models to predict the risk of RILD in a large series of HCC patients treated with SBRT are needed. Moreover, many obstacles to the definition of liver toxicity and meaning of laboratory findings should be overcome in patients with pre-existing liver dysfunction.