Spectroscopic and Multivariate Approaches for Tumor Diagnostics and -Therapy

Abstract

As a result of still increasing cancer incidences worldwide, the demand of novel diagnostics and therapies is more important than ever. The great complexity and heterogeneity of differently originating tumors make it often complicated to apply one overall diagnostic or therapeutic approach. Especially new cancer therapies require a deep knowledge and understanding of the actual operating principal, when applied to certain cancer types for the first time. The highest goals of diagnosis and therapy are to prevent an on-going progression, to improve a patient’s life quality or in best cases to allow a complete cure. To achieve this, new attempts in both fields have to be discovered. One arising possibility in diagnosis are spectroscopy-supported multivariate models for tumor identification and discrimination from normal tissues. For that purpose, various spectroscopic technologies, also applied as imaging modalities, were coupled with multivariate data analysis. Their advantage is to objectively represent chemical or morphological information of tumors or tissues and to use it for their classification into different tumor or tissue types. Moreover, spectroscopy-based multivariate models enable to define reasons for the obtained differentiation and can even be used to predict undetermined tissue areas. In the first part of this thesis, new insights in spectroscopy-based multivariate models for tumor diagnostics are gained. A multi-view approach was chosen to address different aspects in spectroscopy or spectroscopic imaging and multivariate model development. This includes the comparison of Whiskbroom and Pushbroom darkfield elastic light scattering spectroscopic imaging (DF ELSS) models regarding their abilities to identify head and neck tumors in mouse tongues. It was determined, which model is more capable and advanced to diagnose the tumors, also from a practical perspective, and which is additionally more suitable to predict unknown tongue areas. The Pushbroom DF ELSS model was ascertained to be more qualified in lingual head and neck tumor identification and differentiation from gland / muscle and healthy epithelium tissue. Moreover, Pushbroom DF ELSS imaging was defined as the faster high-throughput, easier to handle and less expensive technology for this job. Furthermore, multivariate models on basis of Raman imaging and Fourier-transform infrared (FTIR) spectroscopy were developed for hardly explored salivary gland tumors of the parotid (parotid tumors) in order to differentiate between different tumor entities and to understand the effects of tissue preparation, respectively. The Raman imaging based multivariate model could successfully discriminate two types of benign parotid tumors from normal salivary gland and performed exceptionally in predicting unknown parotid tissues. Multivariate models formed by FTIR data from differently prepared parotid tumors and salivary gland could reveal that formalin fixation had almost no effect on the chemical tissue composition and formalin-fixed tissues closely resembled the native state. Additionally, paraffin and dewaxing treatments greatly impacted the tissue chemistry and in case of dewaxing even caused a loss of component information. All model outcomes were assessed by an additional histopathological characterization and the models themselves were evaluated regarding their quality by performance measures and their practical applicability in clinics. The obtained results should contribute to a transition from laboratory models to real, complementary diagnostic tools in a clinical routine. In the second part of this thesis, new findings were obtained to understand and improve photodynamic therapy (PDT) with hypericin as a new therapeutic treatment for primary brain tumors. For this purpose, in vitro cell culture models were implemented, which represent glioma tumors in the brain. Important aspects of hypericin PDT for glioma treatments are the hypericin’s distribution and accumulation in the tumors and the respective interactions between both under different hypericin incubation conditions. This was studied by fluorescence microscopy and fluorescence lifetime imaging microscopy (FLIM) in a tumor cell spheroid model, which better reflects the 3D cell-cell- and cell-matrix interactions. Fluorescence microscopy exhibited that hypericin gradients into the tumor spheroids occurred at short incubation periods and low hypericin concentrations, whereas long incubation times and high hypericin concentrations caused a homogeneous accumulation across the spheroid. In FLIM analysis, hypericin fluorescence lifetimes (FLT) were related to prevalent, environmental factors in the spheroids. Short incubation periods caused a FLT gradient inside, whereas long incubation times revealed more homogeneous FLTs with partially lower FLTs in the spheroid periphery than in the center. This was correlated to several environmental effects, including pH and metabolic changes, Förster resonance energy transfer (FRET)- or excimer like interactions and cellular redistribution of hypericin. To specifically understand the cell death mechanism for glioma tumors by hypericin PDT, this treatment was applied to glioma cells and analyzed by FLIM in consideration of different irradiation wavelengths. The results showed individual steps of the dying process by using two additional label components. Important steps were cell shrinkage and rounding, bleb formation, medium diffusion into the cell body and membrane bursting. Explanations for each step were deduced and the death mechanism for glioma cells was declared to be initially apoptosis followed by necrosis. With the help of these results, PDT conditions with hypericin can be adapted and its efficiency can be increased to achieve an ideal treatment for brain tumors in vivo.