Beurteilung des Therapieansprechens nach transarterieller Radioembolisation von Lebertumoren mit etablierten Kriterien und Texturparametern

The objective of this dissertation was to determine whether texture parameters of treated liver malignancies have the potential to assess therapy response to transarterial radioembolization (TARE) earlier than by established criteria. Furthermore, we investigated whether texture parameters correlate with time to progression (TTP) or overall survival (OS). So far, only one study using computed tomography (CT), analyzed therapy response to TARE with texture paramaters derived from histogram or texture analysis (PubMed search performed on October 12, 2017 with the following search keywords: Transarterial radioembolization, TARE, selective internal radiotherapy, SIRT or radioembolization combined with histogram analysis or texture analysis). To our knowledge, this is the first study using magnetic resonance imaging (MRI) based texture analyses to analyze therapy response to TARE. In this study, we assessed established sized based response to TARE in 55 cases by means of Response Evalutaion Criteria in Solid Tumors (RECIST) 1.1 and modified RECIST (mRECIST). Response was studied with mint LesionTM including the determination of the following parameters: mean and standard deviation of the intensity values and skewness, kurtosis, entropy and uniformity of the intensity value distribution. Response assessment was performed by delineating the regions of interest on the section with the largest diameter and the best reproducibility. The investigated MRI data contained T1-weighted (T1w), T1w dynamic contrast- enhanced (DCE) MRI and diffusion weighted MRI. For the assessment of response by RECIST 1.1, hepatocellular phase MRI with a hepatocyte specific contrast agent was used. Alternatively, the DCE-MRI phase, in which the liver tumors were best reproducible, was used. In order to obtain the texture parameters in all sequences, the tumors were in addition delineated within the other sequences. For the assessment of response by mRECIST, the arterial phase of DCE-MRI was used. The texture parameters and the changes of the parameters between the phases of DCE-MRI were tested on significant differences between therapy response to TARE by RECIST 1.1 and mRECIST. The texture parameters of baseline MRI were tested on differences between responders and non-responders. We also tested the parameters of the previous follow-up to therapy response on differences between cases with and without progressive disease. Furthermore, we tested whether the texture parameters and the changes of the parameters correlated with TTP and OS. On top of that, we tested whether OS correlated with therapy response to TARE. The analysis was performed for all 55 cases, for cases with primary liver tumors, liver metastases, hypervascular liver tumors and for cases with hypovascular liver tumors, respectively. The results prove the capability to assess and estimate response to TARE, TTP and OS by means of texture parameters earlier than by currently established size criteria. The OS by Kaplan-Meier also correlated with therapy response by RECIST 1.1 and mRECIST. Despite the significant results, limitations shall be considered. The number of cases in this study was limited to 55 cases and the different tumor entities were heterogeneous. To establish texture analysis as a radiomics tool, studies with a larger number of patients and subsequent analyses of the various tumor entities are necessary. Considering the results of this dissertation such studies seem reasonable. A reproducible assessment of therapy response earlier than by established size criteria, would be clinically relevant to improve personalized medicine. In patients with progressive disease, another transarterial radioembolization, an alternative local therapy or a systemic therapy could be initiated earlier.