External beam radiotherapy for unresectable hepatocellular carcinoma, an international multicenter phase I trial, SAKK 77/07 and SASL 26

Patients with unresectable, histologically or radiologically confirmed stage cT1–4, cN0–1 HCC, Barcelona clinic liver cancer (BCLC) stage B and C with no prior malignancy within 5 years were eligible for this phase I trial. Patients had to be older than 18 years, with a Child-Pugh A or B score, with a residual liver volume (= total liver volume – the gross tumor volume (GTV)) of >800 ml and ≥40% of uninvolved liver and WHO performance status 0–2. Metastatic HCC was allowed if ≥90% of total tumor volume was located within the liver. Exclusion criteria included, ALT and AST ≥5x upper limit of normal (ULN), AP ≥ 5 x ULN, bilirubin ≥3 x ULN, hemoglobin ≤ 100 g/L, neutrophils ≤1.2 x 10⁹/L, platelets ≤ 50 x 10⁹/L, international normalized ratio (INR) >2, creatinine clearance ≤ 50 mL/min, clinical ascites, encephalopathy, active hepatitis, gastric, duodenal, or variceal bleeding or weight loss ≥ 15% within three months of registration or esophageal varices ≥ grade 3 [7, 8]. Patients were also excluded, if they had prior RT to the abdomen or caudal chest below T5, prior transarterial chemoembolization (TACE) and radiofrequency ablation (RFA) within eight weeks. No chemotherapy was permitted within three weeks before registration. Portal vein thrombosis (PVTT) was not an exclusion criterion.

The study was planned and conducted in accordance with the principles of the Declaration of Helsinki. The protocol was approved by the ethics committee of each participating site (ClinicalTrials.gov identifier NCT00777894, registered October 21st, 2008). Written informed consent was obtained from all patients.

Pretreatment triphasic diagnostic computed tomography (CT) was used to delineate the gross tumor volume (GTV) and enhanced area of vessel thrombosis. Patients were treated in supine position. Respiratory gating or breath holding techniques were used. Triphasic CT simulation with intravenous contrast was used with a 3- to 5-mm-slice thickness. The range of the simulation CT scan included the whole liver, lower part of the lungs, and both kidneys. The GTV was defined as the visible tumor on the arterial phase of simulation CT and/or fused diagnostic CT. For patients with coexisting PVTT, the area of PVTT was included as a part of the GTV. The clinical target volume (CTV) encompassed the area of the GTV with a 5–10 mm margin. The CTV was expanded by a 5- to 10-mm radial margin and a 6- to 15-mm cranio-caudal margin to create the planning target volume (PTV). Normal liver was defined as the whole liver volume minus the PTV. Patients were treated with either 3D-conformal RT (3D-CRT), IMRT or fractionated stereotactic RT techniques, using 6–18 MV photon energies. Due to compatibility between centers, dose prescription and normalization were fixed at isocenter and for conformal therapy, according to the international commission on radiation units and measurements (ICRU) reports 50 and 62. In case of IMRT, dose prescription and normalization could be defined as mean dose to PTV for optimization reasons, but for reporting it was rescaled to the isocenter. No normalization/prescription to isodose levels was allowed. For 3D-conformal planning: 2 to 5 portal beams with planar or non-coplanar arrangement were used. Larger number of fields was allowed to use to improve the quality of dose distribution [9]. For adequate target coverage, the PTV received 95–107% of the prescribed total dose. Alternatively, a minimum of 95% of the prescribed dose had to encompass more than 99% of the PTV. In addition, it was recommended that areas receiving more than 105% of the prescribed dose, are kept to <1% of the PTV. Our protocol suggested that participants have >800 mL and > 40% of non-tumor liver and that the mean dose has to be kept < 28Gy in patients with a non-cirrhotic liver and < 24Gy in patients with signs of cirrhotic liver. If the PTV encompassed >66% of the total liver volume, radiotherapy was not initiated. RT was delivered from Monday to Friday in five fractions per week, of 2Gy per fraction to a total dose of 10Gy per week. Normal tissue dose delivery guidelines for bowel, stomach, esophagus (each Dmean < 40Gy, V60Gy <3%), lung (V5Gy <85%, V20Gy <20% and Dmean < 12–15Gy), heart (Dmean < 40Gy), spine (D1% <48Gy) and kidneys (Dmean left kidney <12Gy and if V50% of right kidney irradiated >20Gy, Dmean left kidney <5Gy) were provided to facilitate planning [10, 11].

Patients were to be enrolled onto five dose-escalation levels (54, 58, 62, 66 and 70Gy in 2Gy fractions, corresponding to a biological effective dose (BED) of 162, 174, 186, 198 and 210Gy with an alpha/beta ratio of 10, based on a modified 3 + 3 design. In dose levels 1 to 3, three patients, and in dose level 4 and 5, five patients had to be treated per cohort. Escalation to the next dose level was only permitted once no DLT was established within one month after the end of RT. If toxicity occurred in dose levels 1–3, a minimum of six patients and for dose level 4 and 5, a minimum of ten patients needed to be treated at that level, before escalating to the next level. While waiting until one month after RT, at which time the presence or absence of toxicity was determined, subsequent patients could be treated at the pre-defined dose level, up to a dose of 44Gy. DLT was specifically defined for 17 clinical and nine laboratory parameters as grade ≥3 or ≥4 toxicity according to CTCAE vs. 3. The threshold to declare a dose level as MTD, was defined, as a DLT rate of ≤16.7% in dose levels 1–3, and ≤10% in dose levels 4–5. A dose of 54Gy was chosen as the lowest dose level, because doses up to 54Gy showed response rates at around 50% [12].

Before starting patient accrual, each participating center was required to successfully participate in a radiotherapy specific quality assurance (RT-QA) program corresponding to EORTC QART [13] levels 1 and 2: 1.) Facility questionnaire External dosimetry audit (EDA), 2.) dummy run (DR). For the DR an anonymized case including CT data set was made available. Target volumes (GTV, CTV, PTV) and all organs at risk (OAR) were delineated by the center’s investigator and reviewed by the coordinating investigator. Deviations from the protocol were communicated to the participating center, and a revised version of the delineated structures was requested and reviewed. The approved structures were then used to elaborate a treatment plan using the RT technique chosen by the center. The plans were reviewed by the trial chair and trial medical physicist, which requested a revised version, in case of deviations from the protocol.

An internal report comparing the anonymized DR results of the first five participating centers was made available to those centers. Coincidence histograms [14] were used for a quantitative topological comparison of the delineated structures.

Patients were assessed weekly during RT, and after completion of treatment, followed up for one year at 1, 2, 3, 5, 8 and 11 months. At the first month follow-up physical examination, hemoglobin, neutrophils, platelets, hepatic function, renal function, pancreatic lipase, INR and CPS were reviewed. The reason for including pancreatic lipase in the follow up examination was to discover pancreatitis. Liver triphasic CT was performed at 2 and 5 months after RT, thereafter every 3 months until progression. At each follow-up within the first three months after RT, toxicity was graded using the CTCAE vs. 3. RILD was defined as the development of nonmalignant ascites without disease progression and an anicteric elevation of alkaline phosphatase level by at least twofold. Non – classic RILD was defined as the development of jaundice and/or elevated serum transaminases (>5 x UL) within 3 months of completion of RT in patients with underlying chronic hepatic disease (cirrhosis or viral hepatitis). Best objective response of target liver lesions was assessed using RECIST.

Using a modified 3 + 3 design, the planned sample size laid between two and 38 evaluable patients in case of two DLTs in the first two patients and two full cohorts at each dose level, respectively. For the primary endpoint, the following patients were considered evaluable: either patients who experienced a DLT and received a dose of at least 12Gy or patients who completed the treatment according to their dose level. Non-evaluable patients were to be replaced. Primary endpoint was DLT. Best objective response of target liver lesions and adverse events were assessed as secondary endpoints. Local control was defined as time from registration to progression of target lesions or death due to progression, whichever occurred first. Overall survival was the time from registration to death from any cause. The rates at one year of these survival times were estimated using the Kaplan-Meier estimator. Point estimates and, if applicable, the corresponding exact 95% confidence interval were calculated for proportions. Median follow-up time was calculated using the reverse Kaplan-Meier method. All data were collected and analyzed at the SAKK coordinating center in Bern, Switzerland. Analyses were performed using SAS 9.2.

From November 2008 to January 2014, 19 patients from five centers within Switzerland and the Netherlands were recruited; six patients at dose level 1, seven patients at dose level 2 and six patients at dose level 3. One patient was not treated at all with RT because of inadequate normal liver volume < 800 ml and therefor not evaluable for both endpoints.

In DL 3, three additional patients were not evaluable for the primary endpoint: two died during treatment one due to cardiac failure and one due to pneumonia (at 26 and 30Gy in 2Gy per fraction, respectively), and one because laboratory values were repeatedly not measured, making it impossible to assess DLT. The remaining 15 patients completed RT as planned, and were evaluable for the primary endpoint. For the secondary endpoints, 18 patients (95%) were evaluable. Median follow up time was 11.8 months. The median age was 68 years (range 45–82 years, inter-quartile range 62–77). The majority of HCC patients had stage cT3 (47%), cN0 (84%) disease, CP-score A5 (58%) and WHO performance status 0 (53%). Alcohol was the most common etiology of the underlying liver cirrhosis (74%). Two patients (11%) had M1 disease. Median longest diameter of largest lesion was 68 mm (range 18–230 mm). Median administered RT dose was 57Gy (range 26–62Gy in 2Gy per fraction). Other baseline patient characteristics can be found in Table 1. Median longest diameter of largest lesion was 70 mm (range 18–185 mm). Median administered RT dose was 57Gy (range 26–62Gy in 2Gy per fraction).

At dose levels 1 and 2, six patients each and at dose levels 3, three patients were evaluable. One (17%) of 6 patients in dose level 1 and one (17%) of six patients in dose level 2 experienced DLTs (lipase > 5xULN and neutrophils <500/μL). Both, dose levels 1 and 2 had a DLT each among the first three patients, requiring an additional three patients at the same dose level as per protocol. However, dose level 3 (62Gy) was completed, with no DLTs in three patients. Two patients died during RT treatment, one patient in dose level 2 due to cardiac failure, and one in dose level 3 due to pneumonia. One patient in dose level 3 died during follow-up, due to cardiac failure. None of the deaths were related to RT treatment. None of the in total evaluated patients showed any signs of RILD. At dose level 1, one patient developed an elevated lipase value grade 4 (Lipase > 5 x ULN) during the first week of RT, which subsequently normalized on laboratory follow-up. A second patient showed an elevated grade 4 bilirubin value (Total bilirubin > 10 x ULN) one month after RT, which normalized, and a third patient showed an isolated elevated AST grade 4 (AST > 20 x ULN) three months after RT, which also normalized during the subsequent analyses. At dose level 2, one patient experienced neutropenia grade 4 (neutrophils <500/μL) in week 6 of RT and at dose level 3, two patients each experienced lymphocytopenia grade 4 (lymphocytes < 200/mm³) during week 5 and 6 as well as during week 6 and at the end of RT, with no clinical impact.

The overall RECIST response rate was 56% (95% CI 31–78%), with 0% complete response (CR), 56% partial response (PR) and 28% stable disease (SD). In three patients (16%), treated at dose level 3, no response (NR) could be determined. Two patients died and one had symptomatic deterioration before their first CT scan after baseline. An overview of response rate according to DL and CP-Score are shown in Tables 2 and 3. The most frequent site of first progression was outside the treated volume. One-year survival and 1-year local control in this small cohort is 61% (95% CI: 35–79%) and 89% (95% CI 43–98%), respectively, which were not endpoints within the trial (Fig. 1).