Mass spectrometry determination of specific tissue states

Abstract

The present invention relates to determining specific tissue states in histological tissue sections and visualizing their spatial distribution based on spatially resolved mass spectrometric signals. The present invention proposes a method for automatically discovering tissue regions with similar mass spectra and further proposes summing the mass spectra of these regions to improve the spectral quality to a level that allows for more confident identification of known markers of tissue degeneration. Regions with similar mass spectra can be interconnected on a large scale, but can also be isolated on a small scale.

Description

[0001] This application is a divisional application based on the Chinese patent application filed on December 14, 2018, with application number 201811533314.2 and invention name “Mass spectrometry determination of specific tissue states”. Technical Field

[0002] The present invention relates to the determination and visualization of the spatial distribution of specific tissue states (eg, tumor tissue degeneration) in histological tissue sections from spatially resolved mass spectrometry signals. Background Art

[0003] See also Figure 1 Spatially resolved mass spectra of thin tissue sections are obtained by mass spectrometry imaging analysis (imaging mass spectrometry, IMS). For this purpose, a thin tissue section (10) of a deep-frozen tissue slice of the organ of interest from a human, animal or even plant individual is produced, for example, using a microtome and placed on a sample slide (30) (step A). ​​When the MALDI method is used for ionization (MALDI = matrix-assisted laser desorption ionization), a matrix (60) that absorbs the laser energy is subsequently applied to the sample surface; this can be done by intermittently spraying a matrix solution (40, 50) onto the thin tissue section (10) (step B); see, for example, DE 10 2006 019 530 B4, GB 2 437 623 B, US Pat. No. 7 667 196 B2; M. Schürenberg. Raster scanning methods have been established for the subsequent mass spectrometry acquisition.

[0004] The raster scanning method involves scanning a thin tissue section with a focused laser beam pulse (70) in a MALDI mass spectrometer to generate a one-dimensional or two-dimensional intensity distribution for the tissue components, which can be detected in a mass spectrum (200) (step C); see, for example, US Pat. No. 5,808,300; F. Caprioli. Thus, each raster point is irradiated at least once with a finely focused laser pulse (70), and the raster point provides a mass spectrum that can cover a wide range of molecular weights, for example, from 1 to 30 kilodaltons (1 dalton = 1 atomic mass unit u). In modern time-of-flight mass spectrometers, the diameter of the raster point is only about 5 micrometers; up to 10,000 mass spectra per second (200) can be obtained. Fine raster points with a diameter of about 5 micrometers increase the ion yield (80) per analyte molecule and are therefore preferred in principle. However, since the material of the thin tissue section is depleted by the continuous laser irradiation, only a very small number of mass spectra (200) can be obtained from such fine raster points.

[0005] Methods for improving spectral quality include dividing the tissue surface into slightly larger pixels and generating a better quality mass spectrum for each of these pixels. For example, pixels can be sampled in a raster scan and all mass spectra from that pixel can be summed, regardless of tissue type. Or the mass spectrum of a small area (i.e., pixel area) obtained by fine raster scanning can be simply summed. However, it is also possible to generate a mass spectrum of pixels by a laser spot of pixel size and to obtain a mass spectrum of pixels by many repeated laser irradiations on the same target area, also generating a better quality mass spectrum by summing. Pixels are preferably square so as to completely cover the tissue; their side lengths can be from 10 microns to 200 microns.

[0006] Appropriate software can be used to generate so-called "mass images" from the mass spectrum of a pixel. This involves selecting an ion mass in the spectrum that is characteristic of a peptide or protein, or a small mass range around this mass, and graphically displaying its intensity distribution over an area of ​​a thin tissue section. This makes it possible, for example, to correlate the distribution of neuropeptides in the mouse brain, or the distribution of beta-amyloid peptides in the brain of an animal model of Alzheimer's disease, with specific morphological features.

[0007] A significant disadvantage of this simple method for high-quality images is that, to date, only a small number of relatively high-concentration features have been analytically identified in such spectra, such as peptides with high concentrations that are particularly typical for a specific tissue state in a tissue sample. This approach has so far limited the method and prevented its wider application to tissue states that cannot be attributed to the presence of a single peptide or protein.

[0008] Independent of these types of imaging methods, the targeted search for "markers" has developed into an interesting area of ​​clinically oriented research (W. Pusch et al., Pharmacogenetics 2003; 4463-476). This usually involves extracts of body fluids (e.g., blood, urine or spinal fluid) and tissues, which are processed into a crude fraction with less complex analyte components by chromatographic extraction, solid phase extraction or other selective methods and then characterized by mass spectrometry. The mass spectra thus obtained show a more or less complex pattern of mass signals originating from peptides and proteins. By comparing the mass spectra of samples from healthy and diseased individuals, peptides or proteins can be found that (alone or together, sometimes within specific limits of intensity ratio) are characteristic of the individual's health state.

[0009] Figure 2Shown is a suitable representation of the above work by W.Pusch et al. In the first step (left), a set of training data comprising degeneration and normal samples is used to discover biomarker patterns. The clinical diagnoses of all training data samples have been predetermined. Then, bioinformatics programs apply different mathematical algorithms to determine a prediction model based on the biomarker patterns of upregulated or downregulated (even unique) signals. In the second step (middle), the biomarker pattern is evaluated by classifying a set of validation data. Like the training data, this group also comprises degeneration and normal samples with known clinical diagnoses. The prediction model generated in the previous step is used to predict the classification of the sample (normal or degeneration). In the illustration, the analysis is illustrated using the patterns of two biomarkers x and y. Therefore, only two dimensions x and y need to be considered. However, this principle can be easily extended to the patterns of n biomarkers in n-dimensional space. The overall sample obtained is then evaluated by calculating sensitivity and specificity as well as positive and negative predictive values, and this type of prediction / independent verification leads to a better evaluation of these parameters. If the quality of the prediction model is reliable, unknown samples can be classified by the model (step 3, right column).

[0010] Markers obtained in this way can also be used for tissue differentiation by imaging mass spectrometry. Patent specification DE102004037512B4 (US7873478B2; GB2418773B; J. Franzen, M. Schürenberg, D. Suckau) describes how markers can be used for graphical representation of tissue states. This patent specification and all its contents are hereby incorporated by reference.

[0011] Unfortunately, however, it is often the case that the quality of a single mass spectrum from tissue is insufficient to unambiguously detect a marker. The markers are often so weak that they do not stand out sufficiently against the background, or the precision of the signal intensities is so low that additional conditions for the ratio of the different signal intensities of the marker cannot be determined with sufficient precision.

[0012] Purpose of the Invention

[0013] Known markers of tissue state and tissue degeneration cannot usually be detected with sufficient certainty from the mass spectra of individual pixels due to insufficient spectral quality. One object of the present invention is to provide a method that combines the acquired mass spectra to improve the quality, thereby allowing mass spectrometric tissue state differentiation and identification of specific tissue states. Summary of the Invention

[0014] The term "specific tissue state" and, in the extreme case, "tissue degeneration" is understood to mean the state of a region of a tissue section due to stress, pathological changes, infection, changes caused by the influence of xenobiotics, changes caused by genetic engineering, changes caused by mutations, a specific metabolic phenotype, or other changes compared to the normal state of the tissue. Particular emphasis is placed on regions of cancerous tissue.

[0015] The mass spectrum of denatured tissue can differ from that of undenatured tissue, but usually they are very similar. Slight differences are usually not characteristic of variation and therefore cannot be used to identify degeneration. However, the mass spectrum of denatured tissue may contain so-called "markers" that can be used to identify tissue degeneration by mass spectrometry. A marker is a specific characteristic intensity pattern of a substance signal in a mass spectrum (also called a mass signal distribution or mass signal signature) that indicates a specific tissue state. The substances here may be peptides or proteins that are under- or over-expressed to produce a pattern, but they may also include post-translational modifications of proteins, their degradation products, endogenous metabolites, or other substances (such as lipids, glycans, carbohydrates, or even drugs) that accumulate in the tissue. A marker may include a single mass signal or several mass signals. In the case of several mass signals, a condition of signal intensities having a specific ratio to each other can be applied. Markers can be very weak and the characteristic ratio can have very narrow limits: therefore, in order to improve quality, the spectra must be combined so that the signal intensities are presented with sufficient accuracy.

[0016] The present invention proposes a method by which, among other things, tissue regions with similar mass spectra are identified without any other prior information ("unsupervised") and the mass spectra of these regions are summed to improve the spectral quality to such an extent that the probability of confidently identifying known markers of tissue status and tissue degeneration is significantly increased. Regions with similar mass spectra can be interconnected on a large scale, but can also be isolated from one another on a small scale.

[0017] For example, for identity searches in spectral libraries, methods are known for determining the similarity of mass spectra using similarity evaluation coefficients. In the following, mass spectra are considered similar to each other when the comparison of their similarity exceeds a set threshold of the similarity evaluation coefficient.

[0018] The present invention is based on the following experience: denatured tissues often form interconnected regions that show similar mass spectra. However, it is likely that other regions of the same thin tissue section show the same tissue type but are not denatured. One focus of the present invention is to distinguish these regions.

[0019] To find tissue regions with similar mass spectra, the mass spectra of all pixels can be checked against each other for their similarity, but this requires a very large number of similarity determinations. For a small piece of tissue with only about 100*100 pixels (10,000 mass spectra), 5×10 7 If mass spectra similar to the mass spectrum of the selected pixel are excluded from further similarity determinations for other pixels, the number can be reduced to n×(n−1) / 2, which corresponds approximately to half the square of the number of pixels n.

[0020] Furthermore, a method is proposed which requires a plurality of similarity determinations which are only slightly greater than the number of pixels n. To this end, a plurality of pixels ("random seeds") are randomly selected and the mass spectrum of each random seed is now compared with the mass spectrum of pixels from the surrounding environment. As long as the similarity of the mass spectrum to the original random seed is above a specified threshold or greater than the similarity to the adjacent random seeds, the comparison with the mass spectrum of the surrounding environment can be continued in all directions. In this way, an initial interconnected region of pixels with similar mass spectra is found, i.e., regions belonging to the same tissue type, whether degenerated or not, as all experience has shown. The calculation can (but is not necessarily) be terminated when the number of pixels in the region reaches a set value. Pixels that were not assigned to any region after the first run are assigned by repeating the method. A new random seed is then selected from the pixels that have not yet been assigned. The method can be repeated if necessary until the region of interest of the thin tissue section is essentially covered.

[0021] For the presence of markers, each summed mass spectrum from the interconnected region (or only from a portion of the region) is checked. Thus, regions that clearly contain markers are determined to belong to a specific tissue state. The corresponding pixels or positions on the tissue surface can then be marked accordingly in the mass spectrum image. If several regions contain ambiguous marker indications, their mass spectra can be summed so that the improved quality increases the certainty of marker identification. In this case, the sum spectrum can also be used to determine the characteristics of the tissue state, which are then assigned to individual pixels or regions on the tissue slice.

[0022] Before conducting a marker search, it can also be determined in each case whether there is a tissue type for which the marker is actually suitable. Some markers are specific to only one or only a few tissue types. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention may be better understood by referring to the following description, wherein the elements shown are not necessarily drawn to scale but are primarily intended to illustrate the principles of the invention (primarily schematic).

[0024] Figure 1 is a schematic example of a mass spectrometry imaging method by means of MALDI time-of-flight mass spectrometry.

[0025] Figure 2 Schematic diagram of the discovery and use of biomarker patterns for mass spectrometric identification of normal and denatured states.

[0026] Figure 3 Schematic thin tissue sections of a mouse brain with different tissue regions (1) to (6). After summing all mass spectra of region (2), the resulting enhanced spectral quality means that this region can be identified as degenerated with the help of known markers (spectra on the left), while marker detection on the mass spectra of individual pixels does not allow a reliable assignment (spectra on the right). DETAILED DESCRIPTION

[0027] While the invention has been shown and described with reference to several embodiments, those skilled in the art will recognize that various changes in form and details may be made therein without departing from the scope of the technical teachings defined in the appended patent claims.

[0028] It is again pointed out here that the terms "specific tissue state" and "tissue degeneration" are understood to mean the state of a region of a tissue section according to stress, pathological changes, infection, changes caused by the influence of xenobiotics, changes caused by genetic engineering, changes caused by mutations, specific metabolic phenotypes or other changes compared to the normal state of the tissue. For a specific tissue state, it should be known (e.g., by Figure 2 A marker (determined by a method) forms a defined characteristic intensity pattern of a mass signal in the mass spectrum of the tissue. This marker may comprise a single mass signal or several mass signals. In the case of several mass signals, the condition that the signal intensities (also called strengths or amplitudes) have a specific ratio to one another can be applied. The marker may be a very weak mass signal, or the characteristic ratio may have very narrow limits: in order to improve the quality, it is then necessary to combine the spectra so that the signal intensities stand out from the background of the spectrum with sufficient accuracy and to express the ratios of the signal intensities relative to one another with sufficient accuracy.

[0029] The present invention is based on the experience that tissues with these specific tissue states often form interconnected regions. However, these regions exhibit similar mass spectra in terms of certain mass signal characteristics or composition of ion species, but these mass spectra are not necessarily identical to the mass signal distribution of the marker.

[0030] As a first embodiment, the present invention proposes a method for automatically ("unsupervised," i.e., without any other a priori information) determining interconnected regions of tissue with similar mass spectra. Among pixels with similar mass spectra, some or all of the mass spectra of these pixels can then be summed to improve the spectral quality to such an extent that the probability of determining the presence or absence of a known marker is significantly increased. It is well known that the accuracy of the signal-to-noise ratio and signal intensity improves with the square root of the number of spectra summed, i.e., for a sum of four spectra divided by 2, for a sum of 16 spectra divided by 4, and so on. Regions of similar mass spectra can be interconnected on a large scale, but can also be isolated from each other on a small scale.

[0031] Figure 3 Depicted is a thin tissue section through a mouse brain consisting of different tissue regions (1) to (6). For example, tissue region (2) can be identified as specific or degenerative (tumor or similar) with the aid of biomarkers x, y due to the enhanced spectral quality after summing the mass spectra of several similar pixels (spectra on the left), whereas despite the presence of biomarkers x, y, the signal-to-noise ratio of biomarkers x, y in the mass spectrum of a single pixel on the right (shaded black in the figure) is too unfavorable to allow a reliable assignment.

[0032] Methods for determining the similarity of mass spectra using an evaluation coefficient for similarity are known, for example when searching for microorganisms by means of a microbial mass spectrum in a library with a reference spectrum. For example, a publication by Jarman et al. (Analytical Chemistry, 72 (6), 2002, 1217-1223: "An Algorithm for Automated Bacterial Identification Using Matrix-Assisted Laser Desorption / Ionization Mass Spectrometry" describes the generation of reference spectra for a library and the calculation method for similarity analysis between the mass spectrum of a sample under investigation and the reference spectrum of the library. However, the average mass, the spread / range around the average mass, the average intensity, the spread around the average intensity, and the percentage frequency of the mass signal occurring in the repeated spectrum must be determined for each reference spectrum, so this method is not preferred.

[0033] For the identification of microorganisms, the signal quality in a mass spectrometer is crucial. The intensity of the signal is only of secondary importance. In contrast, different tissue states in an organism are usually not distinguished by signals found at different masses. More often, signals of the same mass are observed, but their intensities differ. Unfortunately, the signal-to-noise ratio of the signal in a single spectrum is often insufficient for reliable signal detection in a single spectrum (see Figure 3 , spectrum on the right). This problem can be solved, for example, by performing signal detection on an average spectrum, or by determining the signal position from the individual spectra and then determining the intensity value of each individual spectrum at the signal position from the raw data (e.g., determining the average or maximum value via a mass interval around the signal position).

[0034] As a basis for the determination of the similarity of two or more mass spectra, for example, each mass spectrum can be considered as a vector of identity values. This vector can contain all the mass values ​​of the spectra; however, previous feature detection is usually used to select only those mass values ​​for which the signal is actually present above the omnipresent background (so-called "peak picking", see for example DE19808584 C1; GB2334813B; US6288389B1; J. Franzen). All known methods for determining vector similarity can then be used to determine the similarity value. For example, these are the Minkowski distance, or in particular the Euclidean distance and the city-block (or Manhattan) distance, the cosine distance, the Pearson correlation, etc.

[0035] In order to find tissue areas with similar mass spectra, the mass spectra of all n pixels on the tissue section can be checked against each other for their similarity. This process results in a very large number of similarity determinations. For a small piece of tissue with only about 100*100 pixels (10,000 mass spectra), approximately 5×10 7 (50 million) similarity determinations. The required number m is then expressed by the expression m = n × (n-1) / 2, which corresponds roughly to half the square of the number of pixels n. Such a large number of spectral comparisons requires long computing times, which can be achieved with modern computers. However, computing time can be advantageously reduced if the mass spectra of comparison pixels whose similarity to the starting pixel is above a specified threshold are excluded from further similarity determinations.

[0036] In order to further reduce the computation time, a method is also proposed that requires a number of similarity determinations, which is only slightly greater than the number of pixels n. To this end, a number of pixels ("random seeds") are randomly selected, and the mass spectrum of each random seed is now compared with the mass spectrum of pixels from the surrounding environment. As long as the similarity of the mass spectrum to the original random seed is above a specified threshold or greater than the similarity to the adjacent random seeds, the comparison with the mass spectrum from the surrounding environment can be continued in all directions. In this way, an initial interconnected area of ​​pixels with similar mass spectra is found, that is, areas belonging to the same tissue type, whether denatured or not, as shown by all experience.

[0037] When the number of pixels in a region reaches a set value, the calculation can (but does not necessarily) terminate. Pixels not assigned to any region after the first run are assigned by repeating this method. A new random seed is then selected from the pixels that have not yet been assigned. This method can be repeated as often as necessary until the region of interest of the thin tissue section is substantially covered.

[0038] When substantially all regions have been found, mass spectra from different interconnected regions are examined for similarity in order to determine isolated regions of the same tissue.

[0039] A marker of an abnormal tissue state may be specific only to a single tissue type. Before conducting a search for known markers, it is therefore necessary to determine whether there is a tissue type for which one can be certain that the marker is actually appropriate.

[0040] If an area with a tissue type for which a marker can be used is found, the mass spectra from such an area are now summed, at least as long as the quality of the spectra is high enough to determine beyond a reasonable doubt the presence of the marker. If the number of mass spectra in the area is insufficient for a definitive identification, it is generally possible to determine whether the marker is present with a certain probability or not at all. This examination can be performed on all areas of the same tissue type. The mass spectra of those areas where the marker is likely to be present can then be summed. Spectral quality can thus be specifically enhanced until it is determined that the tissue in these areas is normal or degenerated, i.e., in a specific state.

[0041] The result can be that only some areas of a tissue type exhibit markers of an abnormal tissue state, while other areas of the tissue type are in a normal state. However, this situation is often quite unusual because tissue that is infected, stressed, or otherwise different often also differs from normal tissue in its mass spectrum outside of the markers, albeit only slightly, so interconnected tissue areas with similar mass spectra are often in the same state.

[0042] Another embodiment of this basic method is to first determine the area of ​​the same tissue using known methods, and then sum so many mass spectra from these areas so that the spectral quality is high enough to be able to identify known markers. From DE102010009853B4 (T.Alexandrow; corresponding to US8605968B2 and GB2478398B), this method of determining the tissue area by edge-preserving smoothing images of each mass is known. The tissue area can also be determined using optical microscopy. For this reason, the matrix layer can be removed after obtaining the mass spectrum, and thin tissue sections can be stained and examined under a microscope using conventional methods (see SODeininger and A.Walch: DE102008023438B4, EP2124192B1).

[0043] Another embodiment of this basic method uses a reference mass spectrum of a tissue type from a library as a starting point for a search for similar mass spectral regions. A reference spectrum of a tissue type is selected in which known markers are present for the tissue stress being sought. For this reference mass spectrum, a similar mass spectrum is searched for in the tissue. The mass spectra from the tissue region found here can be summed again, as in the above-described method, to achieve such high spectral quality that the known markers become identifiable.

[0044] All described methods for identifying tissues of a particular state begin with the use of a microtome to form thin tissue slices, preferably from deep-frozen tissue slices or tissue blocks embedded in an organic solid material (e.g., paraffin) and chemically fixed (e.g., formalin-fixed). The thin tissue slices are placed on a suitable support. The support can be a glass sample slide, for example, having a transparent but conductive surface coating for a mass spectrometer on its surface. However, other supports, such as metal supports, can also be used. The glass sample slide is particularly suitable for the above-mentioned method of identifying tissue regions under an optical microscope. Depending on the desired substance class, it may be necessary to perform spatially resolved enzymatic reactions on the tissue surface. For this reason, enzymes are usually sprayed on it.

[0045] The thin tissue section is then sprayed with a solution of a suitable matrix material for ionization by matrix-assisted laser desorption. For example, spraying can be performed on a device that moves the sample slide under the nozzle to produce a uniform spray coating. It must be noted here that the liquid running after spraying does not impair the positional accuracy of the material. The most suitable is intermittent spraying with intermediate drying, as shown in DE102006019530B4 (M.Schürenberg; corresponding to US7667196B2 and GB2437623B). In this process, the crystallized matrix material absorbs these substances from the thin section, which can be embedded in the crystal itself or in the grain boundaries between the crystals during crystallization.

[0046] The choice of matrix substance can significantly influence which biomolecules contribute to the signal in the spectrum. For example, proteins can be prepared using 2,5-dihydroxybenzoic acid (DHB) or sinapinic acid (SA), peptides using 4-cyano-4-hydroxycinnamic acid (CHCA), nucleic acids using 3-hydroxypicolinic acid (3-HPA), and sugar-containing structures using DHB or trihydroxyacetophenone for MALDI MS analysis. The sample is then introduced into a mass spectrometer, and a mass spectrum is acquired.

[0047] First, specify the pixel size based on the generally known tissue structure. Here, we assume that we want to study a slice measuring 5 mm x 5 mm from a fine-structure tissue. For example, a pixel size of 50 μm x 50 μm would be suitable for this. The pixel size should be as large as possible, but at the same time, it should overlap the structure boundaries as little as possible.

[0048] Several methods can be used to obtain mass spectra: (i) using a finely focused laser beam that is raster scanned over a single pixel at a time, comprising the tissue-type-independent sum of mass spectra, (ii) acquiring mass spectra using a laser beam whose focus corresponds to the pixel size, comprising the sum of several successively acquired mass spectra from the same pixel area, or (iii) a hybrid of the above methods. Regardless of the method chosen, the important aspect here is that for each pixel a tissue-type-independent sum spectrum can be obtained that is already of high quality at the outset.

[0049] Raster scanning is discussed here only briefly by way of example. Raster scanning involves the pixel-by-pixel acquisition of mass spectra, wherein at each point in the tissue sample, one or preferably several acquisitions of the mass spectrum are performed by a finely focused laser beam. Regardless of the tissue type, the mass spectra of all points of the pixel are summed in order to achieve a higher dynamic measurement range and also to improve the statistics of the mass signal. The diameter of the "spot" corresponds approximately to the diameter of the laser spot, i.e. the diameter of the laser beam on the sample, which can be set by focusing. The solid-state lasers in mass spectrometers allow a focal length as small as about 5 microns; therefore, the mass spectra of 100 raster points can be acquired in one pixel with an edge length of 50 microns. If four mass spectra can be acquired at each raster point before the material of the thin tissue section is exhausted, we obtain the sum of 400 mass spectra from each pixel. The sum spectrum is saved for each pixel. For a square tissue surface with a side length of 5 mm, this means measuring 10,000 sum mass spectra, each summed from 400 individual mass spectra. Using a 10 kHz acquisition rate, the acquisition of 4 million individual mass spectra took approximately 7 minutes.

[0050] The pixels are usually arranged in a square, parallelogram or honeycomb pattern, but of course they can instead follow the specific morphology of the sample, which can be helpful for example for axons in a ganglion that are several millimeters long. The only important aspect here is that the pixel size appropriately matches the size of the area to be analyzed or the expected anatomical structure.

[0051] The ions produced point by point by MALDI can be examined with a mass spectrometer, which uses various mass analyzers. Typically, a time-of-flight analyzer (time-of-flight mass spectrometer; TOF-MS) with or without an ion reflector is used. However, a time-of-flight mass spectrometer with orthogonal ion injection (OTOF-MS) can also be used. Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR) is also being increasingly used.

[0052] After the measurement, a complete mass spectrum of each pixel can then be obtained. The above-described method according to the invention can then be applied to this data to determine areas of denatured tissue. The tissue status can then be graphically displayed on a screen in the usual manner.

[0053] In summary, the present invention proposes a method for searching for tissue degenerations in histological samples and for visualizing their spatial distribution using spatially resolved mass spectra acquired from thin tissue sections. Markers for tissue degeneration are known, but they cannot usually be identified with sufficient certainty in a single mass spectrum. The present invention is characterized by the fact that regions with similar mass spectra are searched for on the tissue surface by similarity determination, and so many mass spectra from similar mass spectral regions are summed that the quality of the summed mass spectrum is sufficient to unambiguously identify the marker(s).

[0054] For similarity determination, a method is preferably used which provides a similarity parameter. Regions of similar mass spectra are then characterized by the fact that the mass spectra have a similarity to each other above a set threshold value for the similarity parameter.

[0055] Preferably, the tissue surface is subdivided into pixels, and the pixel size is chosen to be appropriate for the tissue structure. Preferably, each pixel is sampled using a raster scanning method, and the mass spectra of one pixel are summed to form a spatially resolved mass spectrum of the tissue for subsequent examination of denatured tissue, regardless of tissue type.

[0056] The first method is characterized by the fact that the mass spectrum of each individual pixel is checked for similarity with all mass spectra of other pixels, thereby generating interconnected regions with similar mass spectra. However, this results in a very large number of spectral comparisons. The number of similarity determinations can be reduced if pixels whose mass spectra are similar to the mass spectrum of the original pixel above a similarity threshold are excluded from further similarity determinations relative to the mass spectra of other pixels.

[0057] The second method is characterized by the fact that the search for the first area begins at an arbitrarily selected pixel and the mass spectrum is determined to be similar to those of all neighboring pixels. The search continues in all directions until, in each case, a boundary is reached where the mass spectra are no longer similar. A second area is then searched, starting at a selected pixel outside the first area, and so on, until the tissue region of interest is virtually completely covered by regions with similar mass spectra.

[0058] Furthermore, the present invention proposes a method for searching for tissue degenerations, characterized by the fact that areas with the same type of tissue are found on the tissue surface using known methods (e.g. optical microscopy) and so many mass spectra are summed from these areas that the quality of the summed mass spectra is sufficient to unambiguously identify (one or more) markers.

[0059] Another method for searching for tissue degenerations in histological samples using mass spectra of thin tissue sections obtained with spatial resolution is characterized by the fact that for one tissue type on the tissue surface, areas of this tissue type are identified by measuring the similarity of the local mass spectrum with a reference mass spectrum, and so many mass spectra from these areas are summed that the quality of the summed mass spectrum is sufficient to unambiguously identify the marker(s).

[0060] The spatial distribution of the tissue state can be displayed graphically, for example on a screen, and for this purpose grayscale or false colors can be used as is customary.

[0061] In addition to the MALDI ionization described in detail above, other types of ionization may also be used with the mass spectrometry imaging methods described here, such as desorption electrospray ionization (DESI) or secondary ion formation by primary ion bombardment (secondary ion mass spectrometry, SIMS).

[0062] The present invention has been described above with reference to various specific exemplary embodiments. However, it should be understood that various aspects or details of the described embodiments may be modified without departing from the scope of the present invention. In particular, if this is feasible for a person skilled in the art, the features and measures disclosed in the different embodiments may be combined as needed. Furthermore, the above description is intended to illustrate the present invention only and is not intended to limit the scope of protection, which is limited only by the appended claims, taking into account any equivalents that may exist.

Claims

1. A method for mass spectrometrically identifying a specific tissue state on a tissue surface of a prepared tissue section, comprising: spatially resolved acquisition of mass spectra from the tissue surface; searching for similar areas on the surface of said tissue by comparing mass spectra obtained using mass signals contained therein, wherein the mass spectra are considered as vectors of intensity values ​​which are subjected to a method for determining vector similarity in order to determine a similarity parameter; generating mass spectra and spectra in regions identified as similar using the similarity parameters, performing marker identification on the sum spectrum, wherein one or more markers are indicative of a particular tissue state; and If identification of corresponding markers on the sum spectrum is successful, all regions on the tissue surface having one or more tissue states on which the sum spectrum is based are characterized. The method of claim 1 , wherein the marker has one or more quality signal characteristics specific to the tissue state.

3. The method according to claim 2, wherein: The similarity of the mass spectra is used to determine regions where the similarity parameter exceeds a set threshold.

4. The method according to claim 1, wherein The tissue surface is divided into pixels for spatially resolved acquisition, whereby a pixel size is chosen that allows acquisition of several individual mass spectra, and whereby a pixel and a spectrum are formed for each pixel from these individual mass spectra.

5. The method according to claim 4, wherein The pixel and spectrum of each individual pixel are checked with the pixels and spectra of all other pixels for similarity.

6. The method according to claim 4, wherein: The number of pixel-by-pixel similarity determinations can be reduced if those pixels whose pixels and spectra are similar to those of the original pixel above a similarity threshold are excluded from further similarity determinations with respect to the pixels and spectra of other pixels.

7. The method according to claim 4, wherein: For the search of the first area, starting from an arbitrarily selected pixel ("random seed"), the similarity of the pixel and mass spectrum with those of all neighboring pixels is determined, and the search is continued in all spatial directions until a boundary is reached in each case where the pixels and spectra are no longer sufficiently similar, at which point a second area is then searched, starting from a selected pixel outside the first area already found, and so on, until the tissue surface has been completely processed.

8. The method according to claim 1, wherein The spatially resolved mass spectrum is compared to a reference mass spectrum that includes mass signals characteristic of the tissue type to determine the similarity of the reference mass spectrum to the spatially resolved mass spectrum.

9. The method of claim 1, wherein the vector contains all quality values ​​of the corresponding spectra.

10. The method according to claim 1, wherein Before being subjected to the method for determining vector similarity, the mass spectra are processed using a peak picking algorithm which selects only those mass values ​​in the mass spectrum where the signal actually exists above the omnipresent background.

11. The method according to claim 1, wherein Methods for determining vector similarity are selected from Ming's distance, Euclidean distance, neighborhood (or Manhattan) distance, cosine distance, and Pearson correlation.

Images