Predicting Role of Dosimetric Parameters for Nonclassic Radiation-Induced Liver Disease After Helical Tomotherapy in Child-Pugh Class a Primary Liver Carcinoma Patients


 PurposeThis study sought to analyze the predicting role of dosimetric parameters for nonclassic radiation-induced liver disease (RILD) after helical tomotherapy (HT) in Child-Pugh (CP) class A primary liver carcinoma (PLC) patients.Patients and methodsA total number of 71 CP class A PLC patients treated with HT from June 2011 to June 2015 were retrospectively reviewed. Clinical characteristics and dose-volume histogram (DVH) were recorded, and liver functions were followed up for 4 months after radiotherapy.ResultsIn all, 57 patients (80.3%) were male, and 14 (19.7%) were female, with a median age of 53 years. The mean gross tumor volume (GTV) was 226.8 cm3. A median dose of 55.0 Gy was delivered by HT with a median fraction size of 2.6 Gy. Twelve patients (16.9%) were diagnosed with nonclassic RILD. The mean dose to normal liver (MDTNL) and the percentage of total liver volume receiving more than 25 - 35 Gy irradiations (V25 - V35) were related to nonclassic RILD. MDTNL showed the highest AUC (0.705, p=0.026). The optimal cut-off value of MDTNL was 21.3 Gy with a sensitivity, specificity and accuracy of 83.3%, 62.7% and 67.6%, respectively. The tolerable volume percentages for DVH were less than: V25 of 42.3%, V30 of 33.9%, and V35 of 28.3%.Conclusion﻿This study suggests that MDTNL, V25, V30 and V35 are dosimetric predictors for nonclassic RILD in CP class A PLC patients. MDTNL < 21.3Gy, V25 < 42.3%, V30 < 33.9%, and V35 < 28.3% may be used to optimize HT planning.

Treatment For simulation and treatment, the patients were trained to breathe shallowly. Simulation was performed with enhanced computed tomography (CT) scan, and two additional series of CT scans during inspiration and expiration were obtained to track the motion of the tumors and other internal organs. The intrahepatic tumor with or without portal vein and/or inferior vena cava tumor thrombi were contoured as the gross tumor volume (GTV). The internal target volume (ITV) was de ned as the summation of the GTVs on the inspiratory and expiratory CT images. The planning target volume (PTV) added a margin of 5 mm to the ITV to compensate for daily set-up errors and target motion.
The OARs were the liver, lungs, kidneys, spinal cord, heart, spleen, esophagus, stomach, duodenum and small bowels. The radiation dose was prescribed to the 95% isodose line encompassing the PTV. The HT was designed to deliver a median total dose of 55.0 Gy (range, 45.0-68.0 Gy) with 2.0-6.0 Gy fractions (2.0-2.2 Gy fractions in 14 patients, and 2.3-6.0 Gy fractions in 57 patients). And the treatment was delivered once per day, 5 times a week using Hi-ART system.
De nition and evaluation of RILDEthics approval and consent to participate RILD occurring between 2 weeks and 4 months after completion of RT is separated into ''classic'' and ''nonclassic'' RILD. Classic RILD is de ned as a triad of anicteric hepatomegaly, ascites, and elevated alkaline phosphatase (more than twice the upper limit of normal or baseline value). [19] Nonclassic RILD involves elevated liver transaminases (more than ve times the upper limit of normal or baseline value), or a decline in liver function (measured by a worsening of CP score by 2 or more). [20,21] Dosimetric parameters selection The dosimetric parameters involved in this study were the dose, fraction size, GTV, normal liver volume, mean dose to normal liver (MDTNL), liver volume, mean liver dose (MLD), the percentage of total liver volume receiving more than 5 Gy (V5), 10 Gy (V10), 15 Gy (V15), 20 Gy (V20), 25 Gy (V25), 30 Gy (V30), 35 Gy (V35), 40 Gy (V40), 45 Gy (V45), and 50 Gy (V50) ( Table 2). These parameters were calculated on the basis of DVH.

Results
Of the 71 patients, 3 patients (4.2%) developed both classic and nonclassic RILD, and 9 patients (12.7%) developed nonclassic RILD alone. Among the 12 patients with nonclassic RILD, all had elevated liver transaminases, and 2 (16.7%) had a worsening of CP score by ≥ 2. The interval to the development of nonclassic RILD ranged from 2 weeks to 4 months (median, 3 months) after HT completion.
Univariate binary logistic regression models were used to evaluate the associations between the clinical / dosimetric parameters and nonclassic RILD, and the statistically signi cant variables were displayed ( Table 3). The correlation of clinical parameters with nonclassic RILD was not detected. Of the dosimetric parameters, MDTNL, V25, V30 and V35 were associated with nonclassic RILD. Notes: p value was tested from univariate binary logistical regression model (method: Enter).
The estimated probability curves of nonclassic RILD were presented (Fig. 1). It was shown that the probability of nonclassic RILD was in proportion to MDTNL. We further tested the predictability of each parameter for nonclassic RILD with the ROC curve, and the area under the curve (AUC) showed meaningful results (Table 4

Discussion
HT has facilitated IGRT system using daily CT image guidance and IMRT delivery using continuously rotating 6-MV linear accelerator and dynamically positioned multileaf collimator. [22,23] With the advent of HT, irradiation of an entire huge PLC or multiple targets can be safely performed with sparing the normal OARs. [24] HT is critical in improving local control and providing a survival bene t as a therapeutic option for PLC. [10,25] It is vital to reduce the probability of RILD when a hepatic irradiation therapy plan is designed. Effective predictors for RILD are required to optimize HT planning, while the published data on RILD after HT in PLC is limited do date.
Classic RILD is considered to be the lethal treatment-related complication for PLC patients and it is very rare in the era of HT. MLD as an important risk factor of classic RILD in 3D-CRT has been well demonstrated in previous publications. Dawson et al. [12] suggested that MLD associated with a 5% risk of classic RILD for primary and metastatic liver cancer were 28 Gy and 32 Gy, respectively. The patients with unresectable intrahepatic cancer were treated using 3D-CRT, and no case of RILD occurred when MLD was less than 31 Gy. [13] Nonclassic RILD is much more common than classic type in PLC patients treated with HT. The incidence of nonclassic RILD in our study was 16.9% (12/71). We found that MDTNL of 21.3 Gy was tolerable in HT for CP class A PLC cases: the incidence of nonclassic RILD was 5.0% (2/40) for patients with MDTNL < 21.3 Gy and 32.3% (10/31) for those with MDTNL ≥ 21.3 Gy. MDTNL is the mean dose delivered to the normal liver, whereas MLD is given to the whole liver. The reason why the dose threshold that we recommend is lower than those of previous studies [12,13] might lie in the background of hepatitis B and liver cirrhosis.
Prior reports have shown that RILD concerned with hepatitis B and liver function status. HBV carriers and CP class B signi cantly correlated to great susceptibility to RILD after 3D-CRT: the incidence of RILD was 33.3% for patients with CP class B and 14.7% for patients with CP class A, and RILD was observed in 24.6% of HBV carrier patients, but only in 4.2% of non-carrier patients. [14] An association between pretreatment liver function and the risk of RILD after SBRT has also been reported. The CP class B cirrhosis was found to have great susceptibility to the development of RILD in HCC patients treated with SBRT.
[26] CP score was a signi cant factor related to liver toxicity and RILD could be tolerated by patients with a CP score not more than seven.
[27] In the current study, nonclassic RILD developed more often in HBV carrier patients than in non-carrier patients (19.1% vs. 12.5%), but this difference was not statistically signi cant. Considering that the CP class of enrolled patients was A, the relationship between RILD and liver function status was not detected. We still recommend that it is necessary to monitor liver function carefully during and after HT.
Furthermore, clinical researches have demonstrated that DVH played an important role in the determination of RILD. The percentage of normal liver volume receiving more than 20 Gy (V 20 ) of 48.5% and the percentage of total liver volume receiving more than 30 Gy ( TL V 30Gy ) of 60% were useful dosimetric predictors for RILD in PLC patients after 3D-CRT. [15,16] Due to the difference of dose distribution between HT and 3D-CRT treatment plans,[25] dosimetric parameters of predicting RILD for HT treatment plans are also needed. The percentage of non-target normal liver volume receiving more than 15 Gy (V 15Gy ) < 43.2% and receiving more than a biological equivalent dose (BED) of 20 Gy (V BED20 ) < 40.8% were recommended to reduce the risk of RILD in HCC patients after HT. [28,29] Our results provided that V25, V30 and V35 were associated with nonclassic RILD in the univariate analyses, and these ndings may be implemental to design, assess and optimize HT treatment plans.

Conclusions
In sum, due to RILD is a serious complication, prevention measures are of paramount importance. MDTNL may be a valuable dosimetric predictor for nonclassic RILD risk in CP class A PLC patients after HT. MDTNL < 21.3 Gy, V25 < 42.3%, V30 < 33.9%, and V35 < 28.3% can be used as reference, and these data may provide useful information in clinical practice. However, further investigation is required to validate the extensive applicability and practical signi cance of our observations via a larger number of patients with greater homogeneity.

Declarations
Ethics approval and consent to participate This study was approved by the Medical Ethics Committee of Zhongshan Hospital, and informed consent was obtained from all patients.

Consent for publication Availability of data and materials
Consent for publication was obtained from all patients.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Abbreviations: RILD, radiation-induced liver disease; SD, standard deviation; GTV, gross tumor volume; MDTNL, mean dose to normal liver; MLD, mean liver dose; VX, percentage of total liver volume receiving more than X Gy Notes: p value was tested from univariate binary logistical regression model (method: Enter).

Competing interests
Abbreviations: RILD, radiation-induced liver disease; OR, odds ratio; CI, Con dence Interval; MDTNL, mean dose to normal liver; VX, percentage of total liver volume receiving more than X Gy. Notes: p value was tested from Student's t-test.
Abbreviations: ROC, receiver operating characteristic; AUC, area under the curve; MDTNL, mean dose to normal liver; VX, percentage of total liver volume receiving more than X Gy. and V35.