Effect of stereotactic body radiotherapy on regional metabolic liver function investigated in patients by dynamic [¹⁸F]FDGal PET/CT

Stereotactic body radiotherapy (SBRT) is increasingly used as a local ablative therapy of liver tumors with emerging evidence that it is effective and safe in both primary [1] and secondary liver cancer [2, 3]. Radiotherapy is, however, still underused for liver tumors and radiotherapy treatment planning is conservative with regards to doses and volumes to the liver, mainly because of sparse knowledge about the radiation tolerance of the liver. Based on surgical experience, it is recommended that at least 700 mL liver tissue be spared and receive a maximum total dose of 15 Gy when SBRT is delivered in 3–6 fractions [4]. It is also recommended that the mean radiation dose delivered to the liver do not exceed 15 Gy in patients without parenchymal liver disease and 13 Gy in patients with viral hepatitis and/or cirrhosis [4]. However, a precise dose–response relationship for liver tissue has not been established and the potential of SBRT in treatment of liver tumors is thus not fully utilized.

Galactose is metabolized in the liver by the cytosolic enzyme galactokinase and the galactose elimination capacity test provides a robust measure of metabolic liver function as demonstrated by its high prognostic sensitivity [5‐10]. The fluorine-18 labeled galactose analogue 2-deoxy-2-[¹⁸F]fluoro-d-galactose ([¹⁸F]FDGal) is also metabolized by galactokinase and is used for functional positron emission tomography (PET) studies of regional metabolic liver function [11‐14]. From dynamic PET scans of the liver with intravenous bolus injection of [¹⁸F]FDGal, so-called parametric images of the liver function can be created which provides voxel-by-voxel values of the hepatic systemic clearance of [¹⁸F]FDGal (K_met, mL blood/min/mL liver tissue). K_met is a flow-independent measure of galactokinase capacity that provides an indirect measure of the maximum removal capacity of [¹⁸F]FDGal regionally in the liver [12‐14] with low individual day-to-day variation [15, 16].

The aim of the present study was to quantify the effect of radiotherapy on metabolic liver function measured by dynamic ¹⁸F-FDGal PET/CT and to determine a dose–response relationship between regionally delivered radiation dose and metabolic liver function in patients treated with SBRT of liver tumors. Metabolic recovery was determined after 3 months.

Average time from baseline [¹⁸F]FDGal PET/CT scan to treatment start was 7 (range 5–10) days. Average time from first treatment to the 1-month scan was 40 (range 36–48) days and average time from first treatment to the 3-month scan was 97 (range 86–109) days. Visually, there was a clear decline in K_met of [¹⁸F]FDGal on images after 1 month, as shown in the example in Fig. 1. Three months after SBRT, the changes were not as marked (Fig. 1D) in accordance with the high regeneration capacity of the liver. Figure 1D illustrates the radiation fields and the dose color wash to the CTV and the normal liver tissue.

Figure 2 shows the relative K_met values for each dose-interval after 1 month and 3 months. In three patients, K_met values in regions that received up to 20 Gy were higher than baseline (105–120% of baseline) after 1 month reflecting a metabolic reserve capacity of the liver. In areas receiving doses higher than approximately 20 Gy, K_met decreased to 30–40% of baseline values with increasing doses. After three months, K_met had normalized in low-dose regions (up to 10 Gy) and increased to approximately 50–85% of baseline values in regions receiving higher doses.

Figure 3 shows the mean dose–response relationship between delivered radiation dose and decrease in metabolic liver function after 1 month. As seen, the relationship was linear until approximately 30 Gy and then became more horizontal. By fitting a polynomial curve to data, a mean TD₅₀ of 17.8 Gy was determined with a 95% CI from 12.0 to 25.8 Gy.

Knowledge of the pathophysiological effects of radiotherapy on normal liver tissue surrounding the target tumor is essential for optimizing SBRT of liver tumors. The liver is known to have a large volume-effect as well as a certain metabolic reserve capacity, but the effect of radiotherapy on metabolic function is largely unknown. In the present study, we aimed to determine a dose–response relationship between radiation dose delivered to normal liver tissue by SBRT and metabolic liver function in terms of hepatic metabolic clearance of [¹⁸F]FDGal and found a mean TD₅₀ of 17.8 Gy. This could have clinical potential for the individual SBRT planning.

Our data do not allow us to make a dose-complication rate relationship e.g. for development of radiation-induced liver disease (RILD) which would require a much larger patient cohort. Moreover, whereas RILD is primarily characterized by hepatic venous clotting [20], [¹⁸F]FDGal PET/CT measures the capacity of the galactokinase enzyme in hepatocytes and the dose–response relationship presented here accordingly reflects a direct effect on hepatocyte metabolic function.

Herfarth et al. studied local changes in the hepatic blood perfusion by multi-phase CT-scans in 36 patients treated by single dose SBRT. Based on the radiological patterns reflecting changes in contrast kinetics after contrast enhancement, focal radiation reactions occurred at a median of 1.8 months after SBRT with a median threshold dose of 13.7 Gy (range 8.9–19.2 Gy) [21]. This is in accordance with our results although it should be kept in mind that the Herfarth data reflects the vascular damage and not the metabolic changes. In the present study, the K_met-values do not depend on hepatic blood flow but solely on galactokinase activity [12, 13]. It seems likely, however, that volumes with decreased perfusion also have decreased metabolic function. This is also supported by a dose-dependent decrease in portal venous blood flow after SBRT [22].

There are some limitations to our study. First, the metabolic reserve capacity of the liver is well described, and it has been shown that the galactokinase activity may increase per volume liver tissue after surgery in both rats [23, 24] and humans [25, 26]. In accordance with this, K_met increased in some of the subjects in the present study 1 month after treatment compared to baseline in areas receiving a low radiation dose. This is a potential disadvantage of the [¹⁸F]FDGal PET method, but it remains one of the only methods for reliable quantification of regional metabolic liver function using a validated method. Moreover, the potential effect on the dose–response calculation was corrected for by normalizing the individual data as described in Materials and Methods. It should also be noted that the baseline values are not subject to reserve capacity and the potential use of [¹⁸F]FDGal PET/CT in functional treatment planning (FTP, see below) is thus not affected by this. Second, the number of patients was relatively low, but the pattern of changes and the dose–response changes were very consistent, and we therefore found the number sufficient for a robust conclusion. In addition, the results are in line with our results in a phase I study using the [¹⁸F]FDGal PET/CT method to optimize the SBRT treatment plan of liver metastases [27]. Third, we only included patients with tumors in otherwise healthy liver tissue. i.e. non-cirrhotic livers and did not include patients with cirrhosis of the liver though the frequency of SBRT used in e.g. patients with hepatocellular carcinoma is increasing due to technological advances [28, 29]. Cirrhotic liver tissue is more vulnerable to radiotherapy than healthy liver tissue and the tolerance dose and recovery rate will most likely be lower than what we found in the present study [4].

Furthermore, we have previously shown that the liver function is more heterogeneously distributed in the cirrhotic liver than in the healthy liver [13] which underlines that careful treatment-planning is important in that subgroup of patients. However, a precise relationship between radiation dose and liver function in hepatitis virus carriers and patients with cirrhosis livers should be determined separately. Individual tolerance to radiation treatment in those patients is likely to also depend on other factors such as individual liver impairment, performance status etc. It should be mentioned that the present method using [¹⁸F]FDGal PET/CT for measuring regional metabolic liver function has been validated in both healthy subjects and patients with cirrhosis, and it is thus readily applicable for patients with parenchymal liver disease as well [12‐14].

A fourth limitation of the study is the risk of mismatch in the deformable co-registration of the PET/CT scan at baseline, 1 and 3 months after SBRT and the planning CT. Motion management and the co-registration of images are factors that should be optimized in the future. Introducing fiducial markers as a surrogate marker of tumor has reduced the uncertainty but were not used in the present study.

An important future perspective would be to utilize [¹⁸F]FDGal PET/CT in individualized treatment of patients with primary and secondary cancer of the liver. Adaptation based on scans acquired during SBRT will not be meaningful due to the late tissue response and the short treatment course of SBRT. However, it seems likely that individualized dose prescription or FTP is possible with sparing of the best functioning sub-volumes of the liver on a pretreatment [¹⁸F]FDGal PET/CT during optimization of the treatment planning [27, 30, 31]. Moreover, techniques for liver-sparing are also warranted due to increased radio-sensitivity in some patients treated with the combination of SBRT and systemic therapy [32]. Accordingly, the indications of SBRT for liver tumors could be expanded to patients with reduced liver function and/or a larger tumor burden than possible to treat with SBRT at the present time point by introducing FTP. Including functional imaging such as the present method also allows for more precise predictions of tissue tolerance instead of modeling [33]. A prospective randomized, case-controlled study for evaluation of the effect of including functional imaging in treatment planning on post-SBRT morbidity and survival would be interesting but should preferably be multi-center for enough patients to be enrolled.

It should be noted that the radiochemical production of [¹⁸F]FDGal is similar to that of the common PET tracer 2-deoxy-2-[¹⁸F]fluoro-D-glucose ([¹⁸F]FDG) [17] which allows for a broad application of [¹⁸F]FDGal PET/CT. If onsite production is not possible, the relatively long half-life of the tracer allows for regional distribution from one center to surrounding centers, like for [¹⁸F]FDG.

We have demonstrated that [¹⁸F]FDGal PET/CT of the liver can be used to define a radiation dose–response relationship between regional metabolic liver function and radiation dose in patients undergoing SBRT for liver tumors and for evaluation of metabolic recovery. The mean TD₅₀ was 17.8 Gy with a linear dose-relationship up to approximately 30 Gy. The development of new methods, such as functional [¹⁸F]FDGal PET/CT used in the present study, is a promising tool for improving the personalized treatment planning by including variation in regional metabolic liver function (inverse dose-painting), thereby sparing the best functioning sub-volumes of the liver and thereby potentially reduce the risk of RILD.