Recognising cancer diseases at an early stage

Researchers at the Ruhr-Universität Bochum (RUB) have developed a new spectroscopic method to support pathologists in diagnosing cancer. They compared conventional procedures for colon cancer identification with a novel method called label-free spectral histopathology. ‘Contrary to previous methods we no longer have to stain the tissue in order to detect cancer,’ says Professor Klaus Gerwert from the Protein Research Unit Ruhr within Europe (PURE) at the RUB. ‘In the future, this will give us the opportunity to classify a tissue sample automatically as being either normal or diseased.’
Today pathologists slice tissue obtained from biopsies into thin sections, stain them chemically, and eventually identify colon cancer by visual inspection under the microscope. This is usually done at an advanced stage of the disease, and the method provides no information about the molecular causes of the tumour. However, the method of spectral histopathology (SHP) established at the RUB Department of Biophysics captures molecular alterations directly in the tissues, especially changes of proteins. It works without any labelling agents, such as fluorescent dyes. SHP may even detect alterations occurring in early tumour stages. Since the analysis uses light beams, SHP is not limited to thin sections of biopsy specimens – in fact, one can apply the method directly in live tissue, where it allows to inspect a site of interest with the aid of fibre-optics. ‘In the future, we intend to work together with clinical partners and apply spectral histopathology to patients directly via endoscopes,’ says Klaus Gerwert.
In SHP, researchers record spatially resolved vibration spectra of a tissue using either an infrared or a Raman microscope. A vibration spectrum reflects the condition of all proteins in a tissue at the site measured. Alterations induced by cancer are reflected in the respective spectrum. The spectrum is thus representative of the status of the sample, just like a fingerprint is characteristic of an individual person. Approximately ten million infrared spectra are usually recorded to produce one single tissue image. Using sophisticated computational image analysis procedures, researchers compare these spectra with a reference database. This database contains spectra of already known tissues and tumours, and has been established in the PURE consortium as a collaboration between pathologists, biophysicists and bioinformaticians. The analytical programme allocates each spectrum to tissue types that have been stored in the database, represented by a specific colour—just like an offender who can be identified by comparing his fingerprints with previous database entries. This produces a spatially resolved annotated image of the colon tissue section. Both PURE members, Professor Andrea Tannapfel, Director of the Pathology Institute at the RUB, and Professor Axel Mosig, Head of Bioinformatics at the Department of Biophysics, made the essential contributions in creating the database and the evaluation algorithm. By now, the evaluation programme will run on any commercial laptop. RUB