Functional Dose-volume Histogram Combining With Plasma Biomarkers For Predicting RILT In Patients Of NSCLC Treated With LCAHRT

BACKGROUNDRadiation-induced lung toxicity (RILT) is the most common dose-limiting complication of thoracic radiation, which can have a considerable impact on patient morbidity and mortality. Clinically significant RILT usually develops in 13%-37% of patients receiving radical dose radiation therapy for lung cancer. Characteristic clinical features associated with RILT include radiation pneumonitis and pulmonary fibrosis. Radiation pneumonitis usually develops in the first few weeks to months after radiotherapy is initiated. Pulmonary fibrosis is the permanent scarring of lung tissue that occurs more gradually over months to years. Several RILT scoring systems such as the Southwestern Oncology Group (SWOG) scale, the Radiation Therapy Oncology Group (RTOG) scale have been developed and published in the literature. Recently, the National Cancer Institute Common Toxicity Criteria (NCI CTC) scale was most frequently used.OBJECTIVE(1)To investigate the clinical parameters (eg. Age, gender, KPS, smoking history, basic pulmonary function), dose-volume histogram (DVH) parameters, and plasma biomarkers association with radiation-induced lung toxicity (RILT) in non-small cell lung cancer (NSCLC) treated with late-course accelerated hyperfractionated radiotherapy (LCAHRT). (2)To identify if functional dose-volume histogram (FDVH) have advantages over conventional DVH for the prediction of RILT. (3) Further to identify if DVH/FDVH parameters combining with plasma biomarkers could provid better prediction efficiency for RILT.METHODSA total of 57 patients of NSCLC treated with radical three dimensional conformal or intensity-modulated radiotherapy were prospectively recruited and followed-up after treatment. The grade of RILT was evaluated according to the NCI Common Terminology Criteria for Adverse Events (version 3.0). Before treatment, all patients performed with:(1)Single-positron emission computed tomograpy (SPECT) of lung perfusion and pulmonary function test (PFTs) in order to evaluate the basic pulmonary function; (2) The SPECT image was registered with the PET-CT data set, and functional lung (FL) was weighted by SPECT lung perfusion. The functional images were then guided LCAHRT, thus conventional DVH was convert to FDVH; (3) Before and after 4-6 weeks of RT, multiple cytokines, including human interleukin-6, super-oxydation dismutase, and transforming growth factorβ1, et al concentrations were measured with enzyme-linked immunosorbent assays (ELISA). The correlation between clinical parameters, conventional DVH and FDVH parameters, plasma biomarkers and RILT were analyzed.RESULTS(1) RILT did not associated with lung perfusion and ventilation injury classification (P=0.077,0.576). Although almost all PFTs parameters trend to decline in RILT group, there were no statistical significance observed, however, if patients with FVC≤2.36 L, FEV1.0≤1.8 L, and FEF25-75%≤1.95 L/s before RT, the occurrence probability of RILT was increased significantly (P=0.017,0.026,0.016, respectively), and the close correlation was found among them (Pearson correlation 0.600-0.898,P<0.001).(2)Univariate analysis revealed multi-parameters of DVH (V5-V40, MLD, NTCP) and FDVH (FV5-FV50, MPWLD, FNTCP) were all statistically signifcant relative to the development of RILT (all values of P<0.05). Multivariate analysis showed only MLD and FV15 associated with RILT (P=0.001,0.044). Receiver operator characteristic displayed almost all FDVH have larger areas under the curve than DVH although no significant difference was observed (P value ranged from 0.066 to 0.951). By subgroup analysis, FDVH have some advantages over conventional DVH for RILT prediction in subgroup of "pulmonary ventilation function injury" (FVC≤2.36L and/or FEV1.0≤1.80L before RT), especially FV45 and FV50 improved prediction efficiency comparing with their counterparts and reached significance (P=0.024,0.048); whereas in subgroup of "perfusion injury" (the areas of perfusion injury reached to one lobe of lung and more), FDVH colud not provide additional value for the prediction.(3) Besides Cyfra21-1, the concentrations of other cytokines were not significant difference between before and after RT; however, besides SOD, there were significant relationship between before and after RT. The ratio of IL-6 before and after RT was significant difference between group RILT and no-RILT (P=0.043). The ratio of TGF-β1 before and after RT and SOD absolute level after RT were borderline significant between group RILT and no-RILT (P=0.054,0.077). If SOD absolute level after RT defined 56.18U/ml as cut-off value, which could be a reliable predictor, with the predictive sensitivity 80.00%, specificity 66.67%. If combinding with dosimetric parameters, the predictive effency would be improving, the sensitivity, specificity, negative predictive value, and positive predictive value would be reached to 80.00%,90.48%,95.00%, and 66.67%.CONCLUSION(1)It is a important clinical significance for selecting high risk population of RILT by evaluating the fundamental lung function with SPECT lung perfusion examination and pulmonary function tests before RT. (2) FDVH are valuable for predicting RILT with the predictive efficiency trending to improving compared with DVH, especially for the patients with pulmonary ventilation function injury, FDVH improved predictive efficiency with some extent.(3) The concentrations of IL-6 and TGF-β1 expressed for higher after RT for patients with RILT. This changes are much more dramatic than patients with on-RILT. The SOD absolute level after RT can be a reliable predictor. If combing with dosimetric parameters, the multi-parameters could improve predictive efficiency at some extent.