gms | German Medical Science

Objective: Nonlinear optical microscopy (NLOM) offers a series of techniques that have the potential to be applied in vivo, for intraoperative identification of tumor borders and in situ pathology. By exploiting the tissue content of lipids, coherent anti-Stokes Raman scattering (CARS) microscopy enables to discern brain tumors from the healthy nervous tissue. Integration of CARS with two-photon excited fluorescence (TPEF) and second harmonic generation (SHG) adds morphological and compositional details. Multimodal CARS imaging of different human tumor biopsies was performed to assess the ability of the technique to retrieve histopathological relevant information.

Method: A total of 40 brain tumor samples were analyzed. The nonlinear microscopy system is composed by a laser scanning microscope and two picosecond fiber lasers emitting in the near infrared. The system enables simultaneous excitation and acquisition of CARS, TPEF and SHG. Tumor biopsies were either imaged in bulk or cryoprotected to prepare cryosections that were imaged without any further processing or staining. Consecutive sections were assessed by conventional histopathology.

Results: Different types of brain tumors were examined: glioblastoma, astrocytoma, metastases of melanoma, mamma carcinoma, lung, colon, and renal cell carcinoma, cranial nerve- and peripheral nerve-associated neurinoma. The borders of the tumors were localized whenever contained in the biopsy. On cryosections, the decrease of the CARS signal intensity in the tumors was quantified up to 60% (for GBM) with respect the surrounding normal tissue, reflecting the overall lower tumor lipid content. The multimodal imaging of fresh bulk samples revealed the native structure of tumor tissue. The acquisition of CARS allowed to image intracellular and extracellular lipid droplets with dimension 1 to 4 µm in GBM (associated to necrosis), as well as in mamma and in renal metastases. GBM and neurinoma were found to express fibrillar collagen in the extracellular matrix, which was imaged by SHG. Fibrotic tissue was visualized also at the border of renal metastases. In the normal tissue axonal structures and fiber bundles were seen and the three-dimensional morphology of myelin sheaths was visualized.

Conclusions: Multimodal NLOM of different human brain and nerve tumors demonstrated the ability of the technique to visualize tumor structures and discern between normal and neoplastic tissue on native tissue samples.