Discussion
We present the data of an international multicenter phase I study of 18 patients treated with cfRT for locally advanced non-resectable HCC. In contrast to previously published prospective studies [
10,
15‐
18], DLT in the present trial was specifically defined for 17 clinical and nine laboratory parameters as grade ≥3 or ≥4 toxicity (CTCAE vs. 3), to address the safety aspect from a biochemical point of view. Additionally, our study-population had large tumors (median 86 mm, range 18–230 mm) and 20% of patients had a CP B score. Even in such a vulnerable patient collective, the present trial showed that cfRT of 58Gy to HCC is safe in an international multicenter setting. A total dose of 62Gy was delivered to three patients without any sign of clinical relevant increased toxicity. However, the maximum tolerated dose could not be determined due to the early termination of the trial because of patient accrual. The reasons for very slow patient accrual included several competing focal treatment options (e.g. RFA, TACE) in liver tumors and the rapid evolvement of SBRT. RT was well tolerated in this study, and no signs of RILD were observed. The DLTs at dose level 1 (54Gy) and 2 (58Gy) (lipase > 5xULN and neutrophils <500/μL) were not clinically relevant. The increased lipase value occurred in the first week of RT at DL 1 and was most likely not related to RT, since it normalized in the subsequent weeks of treatments, when higher cumulative doses were applied. The decreased granulocyte value may have corresponded to an increased granulocyte consumption, which may be interpreted as an expression of intermittent hepatocyte injury [
19]. During follow-up, there was a spontaneous remission of the neutrophils in the physiological range. Therefore it is questionable to use the lipase and neutrophils as a dose-limiting factor for RT induced liver injury. The isolated bilirubin elevation one month after RT and isolated AST elevation at 3 months after RT were only present at dose level 1. No patient in dose level 2 and 3 had similar laboratory changes, even though higher doses were applied. The grade 4 adverse events were transient, and all patients recovered spontaneously within a few months. None of the two patient’s deaths in this trial were related to RT treatment.
Several studies showed that it is safe to treat HCC patients with cfRT in HCC. In a series of prospective trials [
10,
15‐
18], the University of Michigan group first established the safety of an individualized dose allocation approach for liver cancer. They developed a NTCP model that quantitatively described the relationship between dose and volumes irradiated and the probability of developing classic RILD using conformal RT techniques. Radiation dose was individualized based on the volume of normal liver that could be spared without exceeding a 5–20% risk of RILD. Objectively measurable disease was not an entry criterion, although it was followed when available. The prescribed doses ranged from 40–90Gy (median, 60.75Gy) in 1.5Gy twice-daily fractions delivered with concurrent hepatic arterial fluorodeoxyuridine [
2]. In a phase II trial, Ben-Josef et al. [
10] reported median survival of 15.8 months with a trend to improved survival (23.9 vs. 14.9 months) in patients treated with doses of ≥75Gy. Doses below 60Gy had little effect on survival and then a steady increase in survival was observed as RT dose increased to 90Gy. Of the 128 patients, 30% patient developed mostly biochemical grade 3 to 4 toxicity, five patients (4%) developed RILD. In a large retrospective series from Korea [
20] including 158 HCC patients with CPS A or B were treated with 25–60Gy in 1.8Gy daily fractions. The patient selection was similar to the one of the present trial. Median overall survival time was 10 months, with no grade 4 or 5 toxicity reported. They demonstrated that the CP score was a significant factor in the development of RILD and the total radiation dose was the only significant factor determining the tumor response. The same group [
21] reported in a retrospective patterns of care study of 398 patients, with HCC treated at 10 institutions in Korea, that CPS A, tumor size <5 cm, negative lymph nodes and BED > 53.1Gy
(alpha/beta of 10) were significant factors for a better prognosis. In their collective BEDs between 4.2–124.3Gy were delivered, the median survival time was 12 months, and the 2-year overall survival rate was 27.9%.
For that reason we suggest, that dose escalation, exceeding 62Gy, should be chosen based on the NTCP of the surrounding liver tissue. One-year survival and 1-year local control in our trial with this vulnerable cohort were 61% and 89%, respectively, although they were not endpoints in the present study.
The overall RECIST response rate in our trial was 56%, with no patient showing a CR. Mornex et al. [
22] showed in their phase II trial, including 27 patients, a 92% response rate using the WHO and RECIST 1.0 criteria. Ninety-six percent of patients received an RT dose of 66Gy. However, they only included patients with small-size HCC between ≥30 and ≤50 mm, whereas in our trial, tumor size ranged from 18–230 mm. Of all patients, 41% developed grade ≥3 toxicity: 19% asymptomatic grade 3 laboratory parameters toxicities in CPS A patients and 27% grade 4 laboratory parameters toxicities, 15% late grade 3 toxicity consisting of gastric bleeding requiring transfusion, and edematous-ascitic hepatic decompensation requiring paracentesis and diuretics in CPS B patients.
Liu et al. [
23] treated 44 patients with large HCC (60–250 mm) with 40–60Gy in standard fractionation. Tumor response was based on serial CT scans, with an overall response rate of 61% using the WHO response rating criteria. Radiation-induced toxicities remained mild and reversible. Their results are comparable with the results of our study. Similarly, a retrospective study of Toya et al. [
24] treated 38 HCC patients with PVTT and tumor sizes ranging from 9 to 93 mm. A total dose of 17.5–50.4Gy (median 40Gy), in 1.8–4Gy per fraction was delivered, which translated to a BED of 23.4–59.5Gy (median 50.7Gy) with an alpha/beta of 10. Response rate was 44.7%. In 13 patients treated with 45Gy in 3Gy per fraction, the response rate was 76.9%. The PVTT size (≤30 mm vs. ≥ 30 mm) and BED ≥58Gy
(alpha/beta of 10) were factors, which significantly were influencing response rate and survival. The median- and one-year survival was 9.6 months and 39.4%, respectively.
However, using the RECIST criteria to evaluate RT response rate is outdated. It has been shown that extensive tumor necrosis after loco-regional ablative treatment or systemic chemotherapy may not always be followed by an overall reduction in tumor diameter. In some instances the lesion size may even increase due to necrosis [
25]. Several recent studies [
26‐
30] have demonstrated that quantification of residual viable tumor by the European Association for Study of the Liver (EASL) and modified RECIST (mRECIST) guidelines better predict treatment response compared with WHO and RECIST guidelines. Volumetric response assessment is likely to become the gold standard for defining treatment response [
31,
32]. Our protocol was written during a period of time, when the mRECIST guidelines were not standard yet. The available literature on response rates after cfRT reports in WHO or RECIST criteria.
The presented data is well comparable with the existing data within the literature. However, it has its limitations. It is a small number of patients and the maximum tolerated dose could not be determined due to the early termination of the trial. Also the trial duration was long. During this time period other effective treatment techniques such as SBRT or proton beam therapy have evolved. SBRT refers to the use of stereotactic non-coplanar conformal radiation therapy to precisely deliver a large ablative radiation dose in a small number of fractions, while limiting the dose to adjacent normal tissues. The steep dose gradient within the target volume leads to tight conformity with steep and isotropic dose fall-off and high dose delivery to the target volume [
33]. There is a growing SBRT experience, mostly in patients with small (<6 cm) HCC [
34‐
38] with a high local control ranging from 70–90% at one and two years. In a large Canadian phase I/II study by Bujold et al. [
39], 102 patients with locally advanced HCC (median size, 10 cm) were treated with six fractions of SBRT, with a 1-year local control rate of 87% and median OS of 17 months. Despite limiting their study to a CP A score population, CP class deterioration occurred in 29% at 3 months. Proton radiotherapy has also emerged as a treatment option for patients with localized HCC. It enables further dose escalation and precise dose delivery while maintaining a favorable toxicity profile. Various phase II trials have demonstrated the effectiveness and toxicity profile of this therapy [
40‐
42].
Recent studies have compared SBRT and proton beam therapy to other focal treatment options such as RF or TACE [
43‐
45] in early stage HCC patients. Wahl et al. [
43] have published a retrospective study comparing SBRT to RF in inoperable patients with small HCC. For tumors treated with RFA, freedom from local progression (FFLP) at two years was 80.2% vs. 83.8% for SBRT. Increasing tumor size was predictive for FFLP in patients treated with RFA (hazard ratio [HR], 1.54 per cm;
p = 0.006), but not for those treated with SBRT (HR, 1.21 per cm;
p = 0.617). For tumors ≥2 cm, there was decreased FFLP for RFA compared with SBRT (HR, 3.35;
p = 0.025). Takeda et al. [
44] conducted a phase II study, treating 90 CP A and B score patients with a solitary HCC lesion up to a diameter of 4 cm, unsuitable for resection and RF with SBRT and optional TACE. Three-year LC rate and OS was 96.3% and 66.7% (95% CI, 56.3–75.6%) respectively. In an other phase II study Bush et al. [
45] compared proton beam therapy to TACE as a bridge for transplantation. In an interim analysis of 69 subjects, ten TACE and 12 proton patients underwent liver transplantation after treatment. Viable tumor identified in the explanted livers after TACE/proton averaged 2.4 and 0.9 cm, respectively. Pathologic complete response after TACE/proton was 10%/25% (
p = 0.38). The two-year OS for all patients was 59%, with no difference between treatment groups. Median survival time was 30 months (95% CI 20.7–39.3 months). There was a trend toward improved two-year LC (88% vs. 45%,
p = 0.06) and progression-free survival (48% vs. 31%,
p = 0.06) favoring the proton beam treatment group.
However, large, inoperable HCC > 10 cm, remain challenging for treatment, because of close proximity to critical organ, limited liver volume available and a relatively poor liver functional status. In this small niche of treatment indications, when locally ablative treatments like RFA, TACE or SBRT are not possible, cfRT remains a valid treatment approach for liver cancer [
46]. Conventional fractionation schedules may be more robust for certain patients with large tumors or at risk for fibrosis of the biliary ducts. The relatively high alpha/beta ration of 8 [
47] of liver tissue implies highly conformal therapy, if treatment is completed within a few sessions. Nevertheless, the use of SBRT is being preferred whenever feasible. The steep dose gradient within the target volume leads to tight conformity with steep and isotropic dose fall-off and high dose delivery to the target volume and requires, due to the complementary information, when ever available the addition of MRI imaging to GTV delineation as well as appropriate motion control [
48].
Despite increasing utilization, and prospective phase II studies [
39] describing favorable outcomes, SBRT for liver cancer is still not included in practice guidelines [
49‐
51]. There is currently one randomized phase III trial by the RTOG (RTOG 1112, (ClinicalTrials.gov ID:
NCT01730937) open for accrual, comparing Sorafenib versus SBRT followed by Sorafenib in locally advanced HCC. This trial hopefully will help to better clarify the role of RT in HCC.