Radiation-induced liver disease after three-dimensional conformal radiotherapy in patients with hepatocellular carcinoma: dosimetric analysis and implication

Abstract

Purpose: This study is to use the dose-volume analysis to predict radiation-induced liver disease (RILD) for patients with hepatocellular carcinoma (HCC) and a large proportion of chronic hepatitis treated with three-dimensional conformal radiotherapy (3DCRT). Materials and Methods: From November 1993 through December 1999, 93 patients with intra-hepatic malignancies were treated with 3DCRT at our institution. Sixty-eight patients who were diagnosed with HCC and had complete three-dimensional dose-volume data were included in this study. The reasons for patient exclusion were follow-up interval less than 3 months after radiotherapy, incomplete dose-volume data in the palliative treatment, or the diagnosis other than HCC. Fifty of the 68 patients had chronic hepatitis before treatment, either type B or type C. According to Child-Pugh classification for cirrhosis of liver, 53 patients were in class A and 15 in class B. Forty-six patients underwent transcatheter arterial chemoembolization at least a month before 3DCRT to allow adequate recovery of hepatic function. RILD was defined as either anicteric elevation of alkaline phosphatase level of at least twofold and non-malignant ascites (classic RILD), or elevated transaminases of at lease fivefold the upper limit of normal or of pre-treatment level (Grade 3 or 4 hepatic toxicity of Common Toxicity Criteria Version 2.0 by National Cancer Institute)(non-classic RILD). No patient had whole liver irradiation. All patients were evaluated for RILD by physical examination, blood chemistries, and computed tomography of the abdomen, within 4 months from completion of radiotherapy. Three-dimensional treatment planning using dose-volume histogram analysis of the normal liver was used to compare the dosimetric difference between patients with and without RILD. Results: Twelve of the 68 patients developed RILD after 3DCRT, including 2 with classic type and 10 with non-classic type. Mean prescribed dose was 50.2 ± 5.9Gy, in a daily fraction of 1.8-2Gy. None of the patient-related variables were associated with RILD. There was no difference in tumor volume (780ml vs. 737ml, p=0.86), normal liver volume (1210ml vs. 1153ml, p=0.64), V30Gy (percent volume of normal liver with radiation dose >30Gy)(42% vs. 33%, p=0.05), and V50% (percent volume of normal liver with >50% of the isocenter dose)(45% vs. 36%, p=0.06), between patients with and without RILD. Mean hepatic dose was significantly higher in patients with RILD (2504cGy vs. 1965cGy, p=0.02). Patients receiving radiation dose more than 28Gy had a significantly higher rate of RILD (4/5 vs. 8/51, p=0.02). The probability of produced RILD in patients could be expressed as: probability=1/[1+exp-(0.12∗mean dose-4.29)], with coefficients differed from 0 at the level of p= 0.0018 and 0.027, respectively. With the normal tissue complication probability (NTCP) model using volume effect parameter (n) of 0.32, curve steepness parameter (m) of 0.15, and TD50(1) of 40Gy, patients with RILD had significantly higher NTCP than those with no RILD (36.6% vs. 19.1%, p=0.005). With an attempt to estimate a new “n” by grouping patients and using a quartile plot, the best agreement between the calculated and observed complication probability was obtained with the value of 0.28. Conclusion:1 Dose volume histogram analysis can be effectively used to quantify the tolerance of liver to radiation. Patients with RILD had significantly higher mean hepatic dose and NTCP. The hepatic tolerance, expressed as either mean hepatic dose or the parameterization of NTCP model, was much lower in our patients with HCC and a high prevalence of chronic hepatitis, as compared to the other series. Further efforts should be made to test the modified model in an independent clinical trial.