Methods and apparatus for in VIVO molecular characterisation

Abstract

An apparatus for in vivo profiling a region of biological tissue to determine its state of mutation is provided. The apparatus includes a light source configured to generate at least one light beam, an optical arrangement configured to form the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement is further configured to filter the Raman backscattered optical signal to generate a filtered optical signal. The apparatus further comprises an optical detector configured to receive the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and a data processing arrangement configured to process the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

Description

[0001] METHODS AND APPARATUS FOR IN VIVO MOLECULAR CHARACTERISATION

[0002] TECHNICAL FIELD

[0003]

[0001] This disclosure relates to apparatus, methods, and computer-readable medium for in vivo molecular characterisation or profiling of a region of biological tissue to determine its state of mutation.

[0004] BACKGROUND

[0005]

[0002] High grade glioma is an aggressive cancer with an average survival of only 12-14 months. It causes more years of life lost than any other form of cancer1and survival has been unchanged over the past 30 years2. Surgeons currently operate on low and high grade gliomas often without knowing the diagnosis. The current gold standard for surgical operative diagnosis is a frozen section of tissue which takes up to 40 minutes to transport, prepare and report back to the surgeon. This procedure has limited repeatability, is subject to a high degree of inter-observer variability and no molecular information is available. These rare cancers may be composed of many genetic subtypes which cannot be distinguished on routine histological analysis alone3with diagnosis taking up to one week or more after obtaining tissue from surgery. Our understanding of this disease is undergoing a molecular revolution from an imprecise, non-reproducible morphological description to a more robust molecular description as embodied in the 2016 World Health Organisation classification of brain tumours4. Characterisation of gliomas now requires identification of isocitrate dehydrogenase (IDH) mutation. Other key changes such as Alpha Thalassemia / Mental Retardation Syndrome X-linked (ATRX) loss and O6-methylguanine DNA methyltransferase (MGMT) promotor methylation are thought to drive gliomagenesis, confer resistance to treatment and have prognostic and diagnostic value when identified35. Reporting of IDH mutation status can take weeks creating a delay in diagnosis and treatment. There is an unmet need for the rapid molecular identification and characterisation of brain tumours to allow earlier diagnosis. Molecular stratification to match patients with existing or new treatments may also improve outcomes in this devastating disease: one of the goals of precision medicine.

[0006] SUMMARY OF THE DISCLOSURE

[0007]

[0003] According to the present disclosure, there is provided an apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the apparatus comprises a light source configured to generate at least one light beam, an optical arrangement configured to form the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement is further configured to filter the Raman backscattered optical signal to generate a filtered optical signal, an optical detector configured to receive the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and a data processing arrangement configured to process the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0008]

[0004] The apparatus may be configured to determine whether or not the region of biological tissue includes at least one human glioma, tumour and / or cancerous cells based on the state of mutation of the biological tissue. In one example, the state of a mutation may be expressed as a binary output, for example, human glioma present or not present, mutation present or not present, tumour cells / tissue present or not present, and / or cancerous cells present or not present. In one example, the state of mutation may be: normal, infiltrated brain or tumour. In one example, the apparatus may output the state of mutation to the user and / or other apparatus.

[0009]

[0005] In one example, the processing of the detector signal may be to determine the presence of spectral peaks to compute the state of mutation of the biological tissue instead of determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue. In this example, the spectral peaks may be Raman spectral peaks.

[0010]

[0006] The apparatus may be configured to compute the state of mutation in relation to at least one of: an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation, based on the detector signal.

[0011]

[0007] The apparatus may be configured to provide real time in vivo computation of the state of biological tissue to one of: normal, infiltrated brain or tumour.

[0012]

[0008] The apparatus may be configured to provide a real time in vivo computation of the state of biological tissue with a refresh time of less than 10 seconds, optionally less than 2 seconds, or optionally less than 1 second.

[0013]

[0009] The apparatus may be configured to provide real-time in vivo computation of the state of mutation with a refresh time of less than 10 seconds, optionally less than 2 seconds, or optionally less than 1 second.

[0010] The apparatus maybe configured to compute the state of mutation from a Raman spectrum in a range of 300 to 2000 cm-1.

[0014]

[0011] In one example, the at least one of the optical arrangement and one or more algorithms when executed by the data processing arrangement may be configured to reduce an influence of extraneous ambient light radiation when computing the state of mutation, optionally wherein the optical arrangement is configured to reduce the influence of extraneous ambient light radiation by comprising one or more optical filters.

[0015]

[0012] The apparatus may be configured to apply to the detector signal a background correction, a calibration and smoothing, a removal of autofluorescence, to generate corresponding filtered spectrum or spectra for the data processing arrangement to process to compute the state of mutation.

[0016]

[0013] The data processing arrangement may be configured to execute a classification model using at least one of partial least squares computation and linear discriminant analysis, PLS-LDA, to compute the state of mutation in the region of biological tissue.

[0017]

[0014] In one example, the optical arrangement may be configured to include a flexible probe for a user of the apparatus to use to characterise the region of biological tissue, wherein the flexible probe may include a flexible waveguide through which the optical probe beam is guided to the region of biological issue and through which the Raman backscattered optical signal obtained from the region of biological tissue is guided to be processed within the apparatus to compute the state of mutation of the biological tissue, wherein the flexible probed is optionally implemented as a flexible bifurcated optical fiber probe.

[0018]

[0015] The data processing arrangement may be configured to determine a concentration of at least one of following molecular species: glycogen, cholesterol, nucleic acids, amino acids, lipids, (CH2), (CH3), amides, and amide 1.

[0019]

[0016] The optical detector may be implemented as a charge-coupled device, CCD, and / or the light source is implemented as a wavelength-stabilized laser operating to provide a light beam of 785 nm wavelength.

[0020]

[0017] According to another aspect of the present disclosure, there is provided a use of an apparatus according to any examples and / or embodiments described herein, wherein the apparatus is for determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0021]

[0018] According to another aspect of the present disclosure, there is provided a method for an apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the apparatus comprises a light source, optical arrangement, optical detector and a data processing arrangement, the method includes the light source generating a light beam; the optical arrangement forming the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement filtering the Raman backscattered optical signal to generate a filtered optical signal; the optical detector receiving the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and the data processing arrangement processing the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0022]

[0019] According to yet another aspect, there is provided a computer-readable medium having instructions stored thereon which when executed by one or more processors, cause the one or more processors to implement any of the methods described herein.

[0023]

[0020] In the present disclosure, the applicants have developed apparatus and methods as described herein relating to the diagnosis of cancer, such as brain cancer, in vivo. In one embodiment, genetic and epigenetic changes in tissue are predicted or identified in order to guide a surgeon on the tumour removal margin during surgery, predict risk and / or outcome, stratify patients, and / or assist in determining an appropriate treatment plan. The genetic and epigenetic changes predicted or identified according to the present disclosure may be in glial tumours, brain tumours, or any other solid tumours such as breast tumour, lung tumour, bowel tumours or prostate tumour.

[0024]

[0021] The present disclosure may be used for the real-time rapid label-free preparation-free analysis of tissue for the diagnosis of disease or physiological states by detection and measurement of the Raman spectra in a non-invasive manner. That is, the apparatus according to the present disclosure may be used in real-time during surgery, wherein the apparatus gives an indication of presence or no presence of cancerous cells. The cancerous cells may be identified based on isocitrate dehydrogenase (IDH) mutation, Alpha Thalassemia / Mental Retardation Syndrome X- linked (ATRX) loss and / or O6-methylguanine DNA methyltransferase (MGMT) promotor methylation.

[0025]

[0022] In one example, Raman range spectra from 300 to 2000 cm-1are measured. The raw data of each spectrum are background corrected, calibrated, and a smoothing applied. The autofluorescence is removed from each spectrum. Each spectrum is normalised. In a preferred embodiment, Partial Least Squares / Linear Discriminant Analysis (PLS-LDA) classification model is used to predict genetic and epigenetic changes. In an alternative embodiment, Support Vector Machines and Random Forrest classification models may be used to predict genetic and epigenetic changes.

[0026] BRIEF DESCRIPTION OF THE DRAWINGS

[0027]

[0023] Certain examples of the present disclosure will now be described, with reference to the accompanying drawings, in which:

[0028]

[0024] Figure 1 illustrates an example apparatus according to the present disclosure;

[0029]

[0025] Figure 2 illustrates an example method according to the present disclosure;

[0030]

[0026] Figure 3 illustrates an example implementation of an apparatus according to the present disclosure;

[0031]

[0027] Figure 4 illustrates Raman spectrum from collected raw data of brain tissue;

[0032]

[0028] Figures 5a - 5c illustrates tissue site localization with representative in vivo Raman spectrum and biopsy results, respectively, in connection with IDH mutation;

[0033]

[0029] Figures 6a - 6c illustrates tissue site localization with representative in vivo Raman spectrum and biopsy results, respectively, in connection with ATRX mutation;

[0034]

[0030] Figures 7a - 7c illustrates tissue site localization with representative in vivo Raman spectrum and biopsy results, respectively, in connection with MGMT promotor methylation;

[0035]

[0031] Figure 8a illustrates tissue in vivo Raman difference spectra for IDH mutation compared to reference spectra;

[0036]

[0032] Figure 8b illustrates tissue in vivo Raman difference spectra for ATRX mutation compared to reference spectra; and

[0037]

[0033] Figure 8c illustrates in vivo Raman difference spectra for MGMT promotor methylation compared to reference spectra. DETAILED DESCRIPTION OF THE DISCLOSURE

[0038]

[0034] Throughout the following description, specific details are set forth in order to provide a more thorough understanding of the disclosure. However, the disclosure may be practiced without these particulars. In other instances, well known elements have not been shown or described in detail to avoid unnecessarily obscuring the disclosure. Accordingly, the specification and drawings are to be regarded in an illustrative, rather than a restrictive sense.

[0039]

[0035] As used herein, the terms “have,” “may have,” “include,” or “may include” a feature (e.g., a number, function, operation, or a component such as a part) indicate the existence of the feature and do not exclude the existence of other features. Throughout the description and claims of this specification, the words “comprise” and “contain” and variations of them mean “including but not limited to”, and they are not intended to (and do not) exclude other components, integers or steps. Throughout the description and claims of this specification, the singular encompasses the plural unless the context otherwise requires. In particular, where the indefinite article is used, the specification is to be understood as contemplating plurality as well as singularity, unless the context requires otherwise.

[0040]

[0036] The terms as used herein are provided merely to describe some embodiments thereof, but not to limit the scope of other embodiments of the disclosure. It is to be understood that the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise.

[0041]

[0037] The term IDH as used herein should be understood to correspond to Isocitrate Dehydrogenase, the term ATRX to correspond to Alpha Thalassemia / Mental Retardation Syndrome X-linked, and the term MGMT to correspond to O6- methylguanine DNA methyltransferase.

[0042]

[0038] As described in the background section, human gliomas represent a poorly understood heterogenous group of brain tumours with a lack of effective treatments. The present disclosure relate to rapid in vivo molecular profiling using Raman Spectroscopy technology. Raman spectroscopy is a fast, non-destructive, label-free optical laboratory based technique which measures the molecular vibrational modes of chemical bonds. It can be used for analysing biological materials including DNA, lipids and proteins. It has been previously reported as being able to detect cancerous tissue89. However, Raman spectroscopy, as known from the prior art, is a laboratorybased technique that has previously not been used for in vivo molecular profiling of the mutational status of tissue.

[0039] As described in more detail below, the inventors have invented an apparatus using Raman spectroscopy for instant in vivo molecular profiling or characterisation of mutational status of tissue that can be used for identifying cancerous cells, stratifying patients, tailoring treatment plans and / or determining prognosis and / or diagnosis. The inventors have illustrated that an example hand-held probe according to the present disclosure can be used by a surgeon to analyse tissue in real-time during surgery. That is, the apparatus enable precise point Raman measurements of the human brain to be made in the operating room. The inventors have also illustrated that with their invention real time (within 2 seconds) acquisition of in vivo Raman spectrum / spectra from patients undergoing surgery for glioma can be used for instant molecular profiling. In particular, their apparatus can detect in vivo tissue containing IDH mutation, ATRX loss and MGMT methylation with greater than 94% sensitivity and specificity. The identification of these key genetic and epigenetic drivers of tumour formation in real-time is a promising platform for precision medicine in this and other cancers, paving the way for faster diagnostics, rapid patient stratification for targeted treatments and the in vivo profiling of glioma and other complex biological systems. The present disclosure delivers instant advanced chemical analysis using Raman spectroscopy with machine learning which can yield similarly useful molecular information for immediate clinical use.

[0043]

[0040] To this end, the present disclosure provides computing apparatus, methods, and computer program products for in vivo molecular characterisation or profiling of a region of biological tissue to determine its state of mutation. Based on the state of mutation, diagnostics, prognosis, patient stratification and / or targeted treatment plans can be determined. The state of mutation can also be used during surgical removal of a tumour for determining if all of the cancerous tissue has been removed.

[0044]

[0041] Referring now to figure 1 , an apparatus 100 is shown for in vivo profiling, or molecular characterisation of, a region of biological tissue to determine its state of mutation. The apparatus 100 comprises a light source 102 configured to generate at least one light beam 104, an optical arrangement 106 configured to form or process the light beam into an optical probe beam 108 for illuminating or engaging with the region of biological tissue 110 and receiving a Raman backscattered optical signal 112 from the region of biological tissue, wherein the optical arrangement 106 is further configured to filter the Raman backscattered optical signal 112 to generate a filtered optical signal 114, an optical detector 116 configured to receive the filtered optical signal to generate a corresponding detector signal 118 including one or more components indicative of one or more molecular species present in the region of biological tissue. The apparatus further comprises a data processing arrangement 120 configured to process the detector signal to determine from the one or more components an concentration (indication or measurement) of the one or more molecular species to compute the state of mutation of the biological tissue or state of mutation of gene(s) in the biological tissue.

[0045]

[0042] The apparatus may be a probe that can be handheld such that it can easily be maneuvered by the surgeon during surgery. In one example the apparatus comprises more than one component. For example, it may comprise a probe having the optical arrangement 106 that is connected to a data detection and processing unit 122 comprising the optical detector 116 and the data processing arrangement 120. In one example, the optical arrangement 106 is configured to include a flexible probe for a user of the apparatus to use to characterise the region of biological issue, wherein the flexible probe includes a flexible waveguide through which the optical probe beam 108 is guided to the region of biological tissue and through which the Raman backscattered optical signal 112 obtained from the region of biological tissue is guided to be detected by the optical detector 116 and processed by the data processing arrangement 120 to compute the state of mutation of the biological tissue, wherein the flexible probed is optionally implemented as a flexible bifurcated optical fiber probe. In one example, the probe may be detachable from the data detection and processing unit and / or light source 102 by quick release of optical connectors for cleaning and sterilisation.

[0046]

[0043] The apparatus 100, or more specifically, the light source 102, may be operated by a user pressing a switch. The switch may be located on the apparatus 100, or it may be a switch operated by the surgeon’s foot. Alternatively, the apparatus 100 may be operated wirelessly, in that case the apparatus comprises a communication module configured to communicate with a controlling device over a wireless connection such as Bluetooth, Wi-Fi and / or cellular networks including 5G and 6G.

[0047]

[0044] In one example, the apparatus may form part of a robotic system such that it is maneuvered and / or operated by the robotic system.

[0048]

[0045] The apparatus may be operated in real-time such that the surgeon can receive the state of mutation of the biological tissue during surgery. That is, the apparatus may be configured to provide real-time in vivo computation of the state of mutation. This may be with a refresh time of less than 10 seconds, optionally less than 2 seconds, or optionally less than 1 second.

[0049]

[0046] The apparatus may be considered a Raman probe, Raman brain system and / or Raman apparatus.

[0050]

[0047] The optical arrangement 106 may filter the light beam 104 such that the outputted optical probe beam 108 is a monochromatic light such as a laser. In one example, the optical arrangement 106 comprises multiple stages of optical filtering to ensure that the optical probe beam 108 exciting the tissue is of only 785 nm light. In one example, the light source is implemented as a wavelength-stabilized laser operating to provide a light beam of 785 nm wavelength. However, the light source is not limited to this and may be a green, red and / or near-infrared laser. In one example the light beam 104 is a laser.

[0051]

[0048] In one example, the light source 102 and optical arrangement 106 may be considered as a light source or laser source configured to emit the optical probe beam 108 of desired wavelength, e.g. at 785nm wavelength. In another example, the light source 102 and optical arrangement 106 may be replaced with a light source or laser source configured to emit the optical probe beam 108 of desired wavelength, e.g. at 785nm.

[0052]

[0049] As described above, the optical beam probe 108 interacts, excites or engages with the region of the biological tissue. The region of the biological tissue may be considered an area or a part of the biological tissue. The biological tissue may be any tissue of a human body or animal body. In one example, the biological tissue is brain, lung, breast, prostate, abdominal or soft tissue.

[0053]

[0050] Raman backscattered optical signal 112 may be considered the optical probe beam 108 scattered by interaction with the biological tissue 110. The backscattered optical signal 112 comprises a change in the wavelength of light that occurs when the optical probe beam is deflected by molecules in the biological tissue.

[0054]

[0051] As described above, the optical arrangement 106 receives the Raman backscattered optical signal 112 to generate a filtered optical signal 114.

[0055]

[0052] The optical detector 116 receiving and detecting the filtered optical signal 114 may comprise a charge-coupled device (CCD). The optical detector 116 is configured to convert the photons of the filtered optical signal 114 into an electronic charge allowing a part, parts or the entire spectrum to be captured in a single acquisition by dispersing the light from a diffraction grating across an array of pixels on the CCD chip. The optical detector 116 outputs a corresponding detector signal 118 comprising one or more components indicative of one or more molecular species present in the region of biological tissue.

[0053] The components indicative of one or more molecular species may be a spectral feature such as a peak, reading or measurement in a Raman spectrum between 300-2000 cm-1wavenumber indicative of a particular chemical bond, e.g. OH, CH2, CH3, CH4, wherein the component or chemical bond may itself be indicative of a molecular species, e.g. glycogen, cholesterol, nucleic acids, amino acids, lipids, amides, and / or amide 1 present in the region of the biological tissue.

[0056]

[0054] The data processing arrangement 120 receiving the detector signal 118 is configured to determine from the one or more components, e.g. OH, CH2, CH3, CH4, a measurement or concentration of the one or more molecular species. That is, the data processing arrangement 120 may determine the measurement and / or concentration of a molecular species, e.g. glycogen, cholesterol, nucleic acids, amino acids, lipids, amides, and / or amide 1 and based on the concentration of the molecular species compute the state of mutation of the biological tissue. For example, the data processing arrangement may determine the total amount of IDH pg / ml of which a portion of the IDH pg / ml comprises a specific mutation. As such, the data processing arrangement may determine a concentration or measurement of a molecular species of IDH mutation, ATRX loss and / or MGMT methylation in the region of the biological tissue based on the one or more components in the detector signal. In one example, the apparatus is configured to compute the state of mutation in relation to at least one of an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation, based on the detector signal 118. In another example, the apparatus is configured to compute the state of mutation in respect of a measurement of at least one of an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation.

[0057]

[0055] In one example, the processing of the detector signal may be to determine the presence of spectral peaks to compute the state of mutation of the biological tissue instead of determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue. In this example, the spectral peaks may be Raman spectral peaks, e.g. OH, CH2, CH3, CH4 and based on the peaks compute the state of mutation of the biological tissue.

[0058]

[0056] The state of mutation may correspond to the extent of mutation of a gene. For example, the data processing arrangement may compute or predict the status of the IDH gene, the ATRX Retardation Syndrome X- linked gene, and / or the MGMT methylation status. The state of a mutation may be expressed as a binary output, for example, present or not present, mutation present or not present, tumour cells / tissue present or not present, cancer cells / tissue present or not present. The state of MGMT methylation may be expressed as a percentage, for example, 0% or greater than 25% of the tumour cells being methylated. In one example, all of these genes are considered in order compute a comprehensive status of mutation. In one example, the apparatus is configured to compute the state of mutation in respect of a measurement of at least one of: an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation. In one example, the apparatus is configured to predict genetic and epigenetic changes in a region of a biological tissue. In another example, the apparatus is configured for determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0059]

[0057] Mutations in the Isocitrate Dehydrogenase (IDH) genes, specifically IDH1 and IDH2, are linked to several types of cancer, including glioma, acute myeloid leukaemia (AML), and cholangiocarcinoma. These mutations cause a "gain-of-function" where the mutated enzyme converts a- ketoglutarate to an oncometabolite called 2- hydroxyglutarate (2HG), which promotes cancer development by altering cellular metabolism and epigenetics.

[0060]

[0058] ATRX is a protein crucial as a chromatin remodeler that helps maintain genome stability and is frequently mutated in cancers. For example, loss of the ATRX protein in gliomas is associated with a better prognosis, while the retention of ATRX expression in IDH-mutant tumours is linked to a poorer prognosis. ATRX loss may predict IDH / H3F3A mutations and identifies a subgroup of diffuse gliomas with improved survival, making it a useful prognostic marker.

[0061]

[0059] In brain tumours, MGMT methylation indicates a promoter region is turned off, leading to a lack of the MGMT repair protein and better response to chemotherapy like temozolomide. Conversely, an unmethylated MGMT promoter allows the tumour to produce the MGMT protein, which repairs DNA damage and makes the tumour more resistant to temozolomide, resulting in a poorer prognosis with standard treatment. Therefore, MGMT promoter methylation status is a key predictive biomarker for guiding treatment decision.

[0062]

[0060] In one example, the apparatus 100 is configured to determine whether or not the region of biological material includes at least one human glioma, tumour and / or cancerous cells / tissue based on the state of mutation of the biological tissue. That is, the output may be binary, human glioma, tumour and / or cancerous cells / tissue being present or not present. In another example, the apparatus is configured to measure whether or not the region of biological tissue includes at least one human glioma, tumour and / or cancerous cells / tissue.

[0063]

[0061] In one example, the apparatus is configured to compute the state of mutation from a Raman spectrum / spectra in a range of 300 to 2000 cm-1. That is, the components indicative of one or more molecular species may have a wavenumber between 300-2000crrr1, wherein the wavenumber is indicative of a particular chemical bond, e.g. OH, CH2, CH3, CH4.

[0064]

[0062] The data processing arrangement 120 may be configured to execute or use a classification model using at least one of partial least squares computation and linear discriminant analysis (PLS-LDA) to compute the state of mutation in the region of biological tissue. In one example, the PLS-LDA is further used to predict genetic and epigenetic changes in the region of biological tissue.

[0065]

[0063] That is, in a preferred embodiment, Partial Least Squares / Linear Discriminant Analysis (PLS-LDA) classification model is used to determine or predict genetic and epigenetic changes. In an alternative embodiment, Support Vector Machines and Random Forrest classification models may be used to predict genetic and epigenetic changes.

[0066]

[0064] As explained in more detail below, extraneous ambient light can interfere with Raman spectroscopy. The optical arrangement 106 and / or one or more algorithms executed by the data processing arrangement 120 may be implemented in order to reduce an influence of extraneous ambient light radiation when computing the state of mutation. The optical arrangement 106 may include one or more optical filters for reducing the influence of extraneous ambient light radiation.

[0067]

[0065] In one example, the apparatus 100, or more specifically the data processing arrangement 120, may be configured to process the detector signal 118 by applying thereto a background correction, a calibration and smoothing, and / or a removal of autofluorescence, to generate corresponding filtered spectrum / spectra for the data processing arrangement 120 to process to compute the state of mutation. That is, the apparatus 100, or more specifically the data processing arrangement 120, may process the detector signal 118 by applying background correction. This may be achieved by two consecutive spectra being obtained at a preset exposure time typically 1 second. The first spectrum is of a residual ambient light background spectrum with the laser off, and the second spectrum is taken at full laser power (light source 102 being fully on). This may be programmable to allow for different data acquisition conditions. The first spectrum can then be subtracted from the second spectrum to remove noise caused by the ambient light. Autofluorescence of the biological tissue may be removed by applying iterative polynomial fits. The data may also be normalised. These operations may be performed by the data processing arrangement 120. Implementation examples of is shown in figure 4.

[0068]

[0066] Advantageously, the spectra obtained using the apparatus 100 can be done under constant visible illumination and are of high enough quality to go beyond simply recognising abnormal tissue to accurately predict the IDH, ATRX and MGMT status of glial tumours.

[0069]

[0067] Operating the apparatus 100 enables an analysis of the biological tissue in real-time, which takes approximately 2 seconds. It is label and preparation free and readily available within the operative field of the surgeon during the procedure. Through providing molecular information used for diagnosis at the time of surgery, Raman spectroscopy, and more specifically the apparatus according to the present disclosure, can be a valuable tool for the earlier diagnosis of cancer. The information is provided in a timescale useful for surgical decision making and for the selection of appropriate further targeted treatments as early as possible, even at the time of the first operation. The stratification of patients on the basis of molecular information introduces Raman spectroscopy as an important platform for the introduction of precision medicine to cancer. Further clinical trials are being planned to expand the library of Raman brain cancer profiles and evaluate the effectiveness of matching these profiles with existing and novel treatment modalities.

[0070]

[0068] Real-time in vivo detection of key molecular phenotypes in humans offers the opportunity to gain a unique insight into tumour biology. The mapping and studying of gliomas within the exact natural environment of the living patient may represent the most accurate model of this complex disease to understand key drivers of tumour pathogenesis in situ and to better understand tumour biology before, during and after treatment. The inventors have identified tissue containing the most important genetic aberrations driving glioma, but it is likely that Raman spectroscopy will be able to obtain a much more extensive molecular profile of disease in larger multicentre trials. Through further defining human glioma into molecular subgroups and matching these to current and emerging targeted therapies the outlook for this devastating disease can be improved.

[0071]

[0069] Referring now to figure 2, a method 200 for in vivo profiling a region of biological tissue to determine its state of mutation will now be described. The method being performed by an apparatus comprising a light source, optical arrangement, optical detector and a data processing arrangement, the method comprising: the light source generating a light beam; the optical arrangement forming the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, the optical arrangement filtering the Raman backscattered optical signal to generate a filtered optical signal; the optical detector receiving the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and the data processing arrangement processing the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0072]

[0070] The method 200 may comprise any of the features described in relation to the apparatus of figure 1 and / or any of the features of any of the apparatus and methods described herein.

[0073]

[0071] For example, method 200 may comprise determining whether or not the region of biological tissue includes at least one human glioma, tumour and / or cancerous cells based on the state of mutation of the biological tissue.

[0074]

[0072] The method 200 may comprise computing the state of mutation in relation to at least one of: an I DH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation, based on the detector signal.

[0075]

[0073] The method 200 may comprise the data processing arrangement determining the state of mutation when the concentration of the molecular species exceeds a threshold.

[0076]

[0074] The method 200 may comprise providing real-time in vivo computation of the state of mutation with a refresh time of less than 10 seconds, optionally less than 2 seconds, or optionally less than 1 second. The method 200 may comprise computing the state of mutation from a Raman spectrum in a range of 300 to 2000 cm-1.

[0077]

[0075] The method 200 may comprise at least one of the optical arrangement and one or more algorithms when executed by the data processing arrangement reducing an influence of extraneous ambient light radiation when computing the state of mutation, optionally wherein the optical arrangement may reduce the influence of extraneous ambient light radiation by comprising one or more optical filters.

[0078]

[0076] The method 200 may comprise to applying to the detector signal a background correction, a calibration and smoothing, a removal of autofluorescence, to generate corresponding filtered spectrum or spectra for the data processing arrangement to process to compute the state of mutation.

[0079]

[0077] The method 200 may comprise the data processing arrangement executing a classification model using at least one of partial least squares computation and linear discriminant analysis, PLS-LDA, to compute the state of mutation in the region of biological tissue.

[0080]

[0078] In one example, the method 200 may comprise the optical arrangement comprising a flexible probe including a flexible waveguide and the flexible waveguide guiding the optical probe beam to the region of biological issue and receiving the Raman backscattered optical signal obtained from the region of biological tissue and guiding the Raman backscattered optical signal to be processed by the data processing arrangement to compute the state of mutation of the biological tissue, wherein the flexible probed is optionally implemented as a flexible bifurcated optical fiber probe.

[0081]

[0079] The method 200 may comprise the data processing arrangement determining a concentration of at least one of following molecular species: glycogen, cholesterol, nucleic acids, amino acids, lipids, amides, and amide 1.

[0082]

[0080] The method 200 may comprise the optical detector being implemented as a charge-coupled device, CCD, and / or the light source being implemented as a wavelength-stabilized laser operating to provide a light beam of 785 nm wavelength.

[0083]

[0081] A computer product recorded on a machine-readable data storage medium, wherein the software product, when executed on computing hardware, is configured to execute any of the methods described herein.

[0084]

[0082] Example of implementations of the present disclosure will now be described with reference to figures 3 to 8c. The inventors successfully deployed a new intraoperative Raman brain system, or apparatus, shown schematically in Figure 3. The apparatus may correspond to that of figure 1. The apparatus was developed for use in humans during brain surgery. Red (R), yellow (Y), green (G) and blue (B) colouring depicts the relative wavelengths of light being transmitted at various points. Multiple stages of optical filtering were used to ensure only 785 nm light (B) excited the tissue, and only tissue emission at longer wavelengths (RYG) reached the detector.

[0085]

[0083] The steps that were taken to reduce the interference from the operating room lights on the Raman spectra were very effective but still allowed constant visible illumination of the surgical area. This facilitated the accurate placement of the Raman probe during the acquisition of spectra and increased patient safety. Any remaining residual background was reliably removed using the protocol of taking an ambient background light reading first and subtracting this from a subsequent tissue spectrum taken under laser power.

[0086]

[0084] The apparatus was easy to use in a repeatable fashion during each operation performed with spectrum acquisition time being 1 second. The measurement of the calibration standard lamps, taken after all the Raman measurements of each patient were complete, showed small variations in the sensitivity response of the system for different patients. These variations were accounted for in the analyses.

[0087]

[0085] Figure 4 shows an example of a typical calibrated and smoothed raw emission spectrum that was obtained using the developed apparatus, the raw emission spectrum showing Raman peaks superpositioned with a varying level of background tissue autofluorescence. The iterative polynomial fits used to extract Raman emission are shown graphically as dashed curves.

[0088]

[0086] More specifically, Figure 4 shows a “Calibrated and smoothed raw data” curve representing background corrected, calibrated and smoothed raw data with a series of Raman peaks superpositioned on a slowly varying autofluorescence signal. The sharp decline in signal for Raman shifts < 400cm-1was due to optical filters in the system. Dashed curves “Polynomial Fit (Range 394-670cm-1)” and “Polynomial Fit (Range 672-1870cm-1)” represent iterative polynomial fits used to remove the autofluorescence. “Raman (Range 394-670 cm-1)”, in orange, in the bottom left side of figure 4 and “Raman (Range 672-1870 cm-1)”, in blue, in the bottom middle to right side of figure 4 represent the resultant Raman emissions. The labelled Raman peak positions in these two emissions match those found in the literature for common molecular components of tissue.

[0089]

[0087] After the background autofluorescence was removed, a good quality Raman spectrum was obtained for each tissue sample.

[0090] In vivo Data Collection

[0091]

[0088] Raman sampling and tissue biopsy were performed during surgery when deemed safe to do so by the primary surgeons (KO and BV). Figures 5a - 5c, 6a - 6c and 7a - 7c illustrate the chronological process of stereotactic localisation of the tissue sampling point followed by in vivo Raman spectral acquisition according to the present disclosure and finally histological analysis of the subsequent tissue biopsy to validate the findings of the developed apparatus.

[0089] Specifically, Figures 5a-5c, 6a - 6c and 7a - 7c illustrate tissue site localisation with representative in vivo Raman spectrum and biopsy results in relation to IDH mutation, ATRX mutation and MGMT promoter methylation, respectively. The images of Figures 5a, 6a and 7a show sagittal Neuronavigation MRI images showing the sampling point (indicated by an arrow) within the 3D segmented tumour (shown in yellow in coloured produced figures, or in black and white produced figures: highlighted area around sampling point) in tumours. The spectra of Figures 6a, 6b and 6c show average Raman spectra for IDH mutated and non-mutated, ATRX mutated and non-mutated and MGMT methylated tissue compared to non-methylated tissue, respectively. In the spectrum of figure 5b, “IDH WT” represents wild-type IDH which is non-mutated and “IDH MT” represent mutated IDH, in the spectrum of figure 6b “ATRAX Retained” represent ATRX protein expression and “ATRX Loss” represent loss of ATRX protein, and in the spectrum of figure 7b “MGMT methylated” represent methylated MGMT and MGMT unmethylated represent unmethylated MGMT. The spectra in the range from 670 to1870 cm-1have been scaled by 10 x for clarity. The images of Figures 5c, 6c and 7c show corresponding immunohistochemistry pathology slides of tissue samples taken showing presence of IDH1 mutation, loss of ATRX expression and presence MGMT promotor methylation, respectively. Abbreviations used here are: IDH = Isocitrate Dehydrogenase gene, ATRX = Alpha Thalassemia / Mental Retardation Syndrome X-linked gene, MGMT = O6- methylguanine DNA methyltransferase.

[0092]

[0090] For clarity, the average in vivo Raman spectra are shown for each mutational state. Spectra were taken under full filtered microscope illumination, stereotactically logged with a neuro- navigation system (using Sonowand and 3D slicer, pre-operative MRI and intra-operative 3D ultrasound) and matched to tissue samples taken at the exact location Raman spectroscopy was performed. Using our in vivo Raman system 471 Raman spectra were collected from 17 patients undergoing brain surgery for WHO grade 2-4 gliomas (Tables 1 and 2).

[0093]

[0091] Table 1

[0094] WHO Grade Diagnosis Number of Patients

[0095] 2 6

[0096] Astrocytoma 5 Oligodendroglioma 1

[0097] 3 2

[0098] Astrocytoma 1

[0099] Oligodendroglioma 1

[0100] 4 9

[0101] GBM 9

[0102] Age: median (range) 42 (22-79)

[0103] Total17

[0104]

[0092] Table 1 sets out patient demographics and histological diagnosis. That is, 17 patients were enrolled in this study with a range of low and high grade tumours. All tissue samples were subjected to molecular histopathological analysis in line with the World Health Organisation 2016 classification of brain tumours2GBM=Glioblastoma Multiforme.

[0105]

[0093] All Raman spectra came from tumour tissue sites that were stereotactically correlated with matching biopsies. The biopsy samples subsequently had one or more molecular test performed on them (IDH, ATRX, or MGMT) as summarised in Table 2.

[0106]

[0094] Table 2

[0107] Molecular Test Status Number of Patients Number of

[0108] Spectra

[0109] IDH Mutated 10 258

[0110] Wild Type 7 171

[0111] ATRX Loss 7 164

[0112] Retained 9 230

[0113] MGMT Methylated 7 128

[0114] Unmethylated 10 186

[0115] Total 17 471

[0116]

[0095] Table 2 illustrates a summary of IDH, ATRX and MGMT mutational states and the number of spectra collected from tissue for molecular analysis. A total of 471 spectra were collected from 17 patients. IDH= Isocitrate Dehydrogenase gene, ATRX=Alpha Thalassemia / Mental Retardation Syndrome X-linked gene, MGMT=O6- methylguanine DNA methyltransferase.

[0117] Raman Spectra for IDH, ATRX and MGMT

[0118]

[0096] High quality in vivo spectra with clear peaks were obtained for all tumours investigated.

[0119]

[0097] Figures 8a - 8c illustrate in vivo Raman difference spectra for IDH mutation (“ IDH MT - IDH WT” lower spectrum in figure 8a), ATRX mutation (“ATRX Loss- Retained” - lower spectrum in figure 8b) and MGMT promotor methylation (“MGMT Methylated-Unmethylated” - lower spectrum in figure 8c) compared to reference spectra (upper spectra in figures 8a-8c). That is, difference spectra are shown with reference spectra from “2-Hydrogyglutarate” (upper spectrum in figure 8a) and DNA (upper spectra in figures 8b and 8c) that have been shifted on the intensity scale to aid comparison. The red dots (shown as black dots in black and white produced figures) on the difference spectra are locations of significant difference (<0.05) as determined by the Mann Whiney II statistic. 15 out of 192-HG peaks align with peaks in the IDH difference spectrum, and most DNA peaks align with troughs (loss) in the ATRX and MGMT difference spectra suggesting strong associations as indicated by the dashed vertical lines. Abbreviations used: IDH= Isocitrate Dehydrogenase gene, ATRX=Alpha Thalassemia / Mental Retardation Syndrome X-linked gene, MGMT=O6- methylguanine DNA methyltransferase. The difference spectra in the range from 670 to1870 cm-1have been scaled by 10 x for clarity.

[0120]

[0098] Referring again to figures 5a -7c, averaged spectra are shown in figures 6a, 6b and 6c together with indicative immunohistochemistry results for each genetic or epigenetic change and the location of the in vivo Raman spectrum analysis point on the 3D navigational MRI scan. It can be seen that these averaged spectra have unique Raman signatures spanning the DNA, protein and lipid domains of the Raman spectrum. The presence of IDH mutation and MGMT promotor methylation was further confirmed using sequencing for IDH1 and IDH2 mutations and using a PCR assay for MGMT promotor methylation. Difference spectra were calculated to show the effect of genetic and epigenetic changes observed on Raman brain spectra by subtracting the averaged spectra of tissue containing IDH and ATRX mutation or MGMT promotor methylation only from non-mutated or non-methylated tissue. These peaks are compared with reference peaks of 2-hydroxyglutarate (2-HG) and DNA as shown in Figures 8a - 8c. For IDH mutation the resulting difference spectrum matches 15 out of the 19 key peaks seen in the reference spectrum for 2-HG, a metabolite generated exclusively by mutant IDH activity10. There is a good correlation with Raman reference peaks for DNA and the ATRX difference spectrum with a particularly strong correlation of the 790 cm-1peak which relates to DNA backbone vibrational modes. Similarly, MGMT promotor methylated tissue shows a strong correlation with DNA reference peaks, with nearly all DNA peaks being represented in the MGMT difference spectrum.

[0121] Classification Models

[0122]

[0099] Different classification models for predicting genetic and epigenetic changes in tissues were assessed. The best performing model for predicting genetic and epigenetic changes in tissue based on the in vivo Raman spectra collected was Partial Least Squares / Linear Discriminant Analysis (PLS-LDA) with Support Vector Machines and Random Forrest producing slightly inferior results. Over training was avoided by both using a leave one out cross validation (LOOCV) process and limiting the number of PLS components to 12 or less.

[0123]

[0100] Tissue containing the IDH mutation was classified with 94.2% sensitivity and specificity, whereas tissues with ATRX loss were classified with 95.1% sensitivity and 96.1% specificity and those with the MGMT promotor methylation classified with 98.4% sensitivity and 98.9% specificity. The Area Under the Curve (AUC) for the generated Receiver Operating Characteristics (ROC) curves was consistently above 98% showing excellent performance of the PLS-LDA predictive model (Table 3).

[0124]

[0101] Table 3

[0125]

[0102] Table 3 illustrates Partial Least Squares / Linear Discriminant Analysis (PLS- LDA) performance using Raman spectroscopy for the prediction of key genetic and epigenetic changes in glioma. Sensitivity, specificity and Area Under Curve (AUC) for the Receiver Operating Characteristic (ROC) was calculated for each type of change observed. PLS-DA classified IDH and ATRX mutations and MGMT promotor methylation with excellent sensitivity and specificity, and a high AUC, demonstrating excellent classifier performance. Abbreviations used: IDH= Isocitrate Dehydrogenase gene, ATRX=Alpha Thalassemia / Mental Retardation Syndrome X-linked gene, MGMT=O6-methylguanine DNA methyltransferase.

[0126] Discussion

[0127]

[0103] The ability of Raman spectroscopy to detect changes in tissue caused by cancer at a molecular level for in vivo clinical use has been limited by the challenge of the low signal of inelastically scattered light11. The inventors have demonstrated here a methodology capable of high quality in vivo Raman spectroscopy for the human brain comparable to laboratory acquired spectra. Furthermore, the spectra for IDH, ATRX and MGMT mutated tumour tissues are unique. Specific DNA, lipid and protein regions of the Raman spectrum were altered suggesting that changes at multiple biological scales are responsible for the differences observed in glioma. Peak analysis has revealed some tentative underlying mechanisms. The average spectrum from tissues containing the IDH mutation had peaks that correlated well with peaks in the spectrum of 2-HG, a metabolite which is known to accumulate in IDH mutated cells10. The strong DNA peaks attributed to backbone vibrational modes seen in the average spectrum from ATRX mutated tissue correlates well with the known involvement of ATRX in chromatin remodelling: specifically, the binding of the H3.3 histone to telomeric regions of chromosomes12. Dysfunction of this pathway is thought to lead to genomic instability through the ALT phenotype13. The key differences seen in the average spectrum from MGMT promotor methylated tissue compared to nonmethylated tissue are mainly due to the DNA region of the Raman spectrum, which matches the observation that methylation of the MGMT promotor silences MGMT expression and impairs DNA repair causing accumulation of damaged, hypermethylated DNA14.

[0128]

[0104] Raman spectroscopy has already been demonstrated in vivo as being able to distinguish normal tissue from tumour tissue in the human brain15and other organs9. However, the efficacy of Raman spectroscopy to detect distinct molecular phenotypes has been difficult to demonstrate ex v / vo16and in vivo data is lacking in part due to the challenges of obtaining adequate in vivo Raman spectra without fluorescence15. We have shown it is possible to acquire spectra in the clinical environment where ambient background light and signal to noise ratio have previously been problematic. The spectra reported here were obtained under constant visible illumination and are of high enough quality to go beyond simply recognising abnormal tissue to accurately predict the IDH, ATRX and MGMT status of glial tumours.

[0105] This analysis is in real-time, takes about 2 seconds, is label and preparation free and readily available within the operative field of the surgeon during the procedure. Through providing molecular information used for diagnosis at the time of surgery Raman spectroscopy can be a valuable tool for the earlier diagnosis of cancer. The information is provided in a timescale useful for surgical decision making and for the selection of appropriate further targeted treatments as early as possible, even at the time of the first operation17. The stratification of patients on the basis of molecular information introduces Raman spectroscopy as an important platform for the introduction of precision medicine to cancer. Further clinical trials are being planned to expand the library of Raman brain cancer profiles and evaluate the effectiveness of matching these profiles with existing and novel treatment modalities.

[0129]

[0106] Real-time in vivo detection of key molecular phenotypes in humans offers the opportunity to gain a unique insight into tumour biology. The mapping and studying of gliomas within the exact natural environment of the living patient may represent the most accurate model of this complex disease to understand key drivers of tumour pathogenesis in situ and to better understand tumour biology before, during and after treatment. We have identified tissue containing the most important genetic aberrations driving glioma, but it is likely that Raman spectroscopy will be able to obtain a much more extensive molecular profile of disease in larger multicentre trials. Our hope is that through further defining human glioma into molecular subgroups and matching these to current and emerging targeted therapies the outlook for this devastating disease can be improved.

[0130] Method

[0131] Study Design

[0132]

[0107] This prospective observational cohort study was an IDEAL framework18stage 2a (development of an idea looking at technical and procedural success) and early stage 2b (exploring clinical indications of new technology) study based at Imperial College Healthcare NHS Trust, London. The study was approved by the National Research Ethics Service and is registered on the National Institute for Health Research Portfolio ID 16738, and listed under the Cancer Research UK Trials Registry, trial number 12192. Recruitment commenced 14 / 03 / 15 and finished 16 / 08 / 16. The inclusion criteria were all adult patients undergoing craniotomy and resection of brain tumours at our institution. Patients unable to consent and those undergoing emergency surgery were excluded, and patients were free to withdraw at any stage during the study.

[0108] The primary outcome measure for this study was the technical and procedural success in safely recording accurate Raman spectra during surgery in a range of patients and brain tumours. Secondary outcome measures included the sensitivity and specificity of detecting different mutations found in glioma. Adverse event reporting was through the UK Good Clinical Practice research governance framework. Patients were informed that the surgery being performed would not change as a result of being included in the study and biopsies would only be performed when the senior operating surgeon was satisfied that this would be safe to do so.

[0133] Raman System Design, and use in the Operating Room

[0134]

[0109] A commercial Raman system (Verisante AuraTM) was provided by Verisante Technology Inc. Richmond, B.C., Canada and modified for use in this study. A schematic of the Raman system is shown in Figure 3. The design was based on similar systems used for in vivo studies on human skin, lung, colon, and oral cavity19-22. The Raman excitation source was a wavelength stabilised 785 nm class butterfly diode laser pigtailed to a 105 pm diameter optical fibre (PD-LD, NJ, USA) and driven by a Verisante designed control module. Tissue emissions were spectrally analysed with a Verisante, in-house designed, f / 2 spectrograph incorporating a holographic transmission grating. The intensity of the dispersed emission was quantified with a front illuminated, thermo-electrically cooled (-20°C), charge coupled device (CCD) array with 1024 x 1024 pixels (Apogee F6, Andor Technology Ltd, Belfast, UK). The size of each CCD pixel was 24 x 24 pm, resulting in a total chip size of 24.6 x 24.6 mm. The system was capable of measuring Raman spectra with shifts of up to 2000 cm-1at a maximum spectral resolution of 7.7 cm-1for a spectrograph object width of 100 pm.

[0135]

[0110] A specially designed flexible bifurcated fibre optic probe was used for delivering the excitation light to the tissue surface and for collecting the Raman emission from it. The common end was 2.8 mm in diameter and consisted of a centred 200 pm diameter excitation fibre surrounded by 34, 100 pm diameter, collection fibres. The probe could be detached from the rest of the system by quick release optical connectors for cleaning and sterilization. Multiple stages of optical filtering were used within the system. On the emission collection pathway, long pass (LP) interference filters, with optical densities (OD) ranging from 4-6 for Raman shifts < 395 cm-1 (< 810 nm) were used at three locations: collection fibre end face, between the catheter and spectrograph input fibre, and before the grating inside the spectrograph. These LP filters nullify the interfering effects from elastically scattered laser light reaching the CCD directly or from generating interfering light emissions in the optical components of the collection pathway. On the excitation light pathway, both short pass (SP) and band pass (BP) interference laser line (LL) filters were used: one between the laser output fibre and the catheter, the other at the end face of the excitation fibre. These LL filters nullify the interfering effects from light emissions generated in the optical components of the excitation light pathway. Control of the system was implemented by a personal computer (PC) using a custom designed program.

[0136]

[0111] The operating room was prepared for Raman measurements by reducing the ambient light background from all sources including the overhead operating light and the surgical microscope light23. Unnecessary lights were turned off or blocked when turning off was not an option. Other critical light emitting devices were either orientated or optically filtered to prevent significant interfering light emissions (see supplementary information: methods section). A foot switch was used by the clinician to trigger data acquisition. Once triggered, two consecutive spectra were obtained at a preset exposure time typically 1 second. The first was a residual ambient light background spectrum with the laser off, and the second was taken at full laser power, (which was programmable to allow for different data acquisition conditions). In all cases a maximum excitation laser power was determined by assuring that the calculated energy density at the tissue surface was within the American National Standards Institute (ANSI) standard for the maximum permissible exposure limit of skin24. The program then subtracted the residual background spectrum from the spectrum obtained under laser excitation to obtain the raw spectral signal.

[0137]

[0112] The raw spectra were wavelength and intensity calibrated using coefficients determined from the measurement of emission standards and stored electronically in the system2526(see supplementary information: methods section). The correct Raman shift was determined from an accurate measurement of the laser wavelength (784.45 nm). The background and calibration corrected raw data for each spectrum was smoothed with a 5 point moving average, and then an iterative polynomial background subtraction algorithm27was used to remove background autofluorescence. All data corrections occurred within a few hundred milliseconds with the resulting Raman spectrum shown on the PC monitor in real-time similar to that shown in figure 4. The data for each spectrum was automatically saved for offline analyses.

[0138] Study Protocol

[0139]

[0113] All participants in this study underwent tumour resection through an open surgical technique (standard craniotomy). At various stages during the operation Raman analysis of tissue was performed using the sterile handheld Raman probe, apparatus, attached 5 mm from the tip of a navigation probe positioned over the brain tissue being analysed. Positioning was performed using an operating microscope (Leica M530 OH6) which was optically filtered to remove all interfering near IR emission. The microscope optical filtering, and other measures led to a virtual elimination of the contaminating ambient background light illumination. Under these conditions Raman spectra were systematically obtained over the normal cortical surface, on route to tumour, superficial and deep tumour zones and the resection margin representing the end point of surgery. The 3D neuro-navigation position (based on preoperative MRI and intraoperative 3D ultrasound) and a matched core biopsy for histological analysis was taken at each reading site using Sonowand Neuronavigation and 3D Slicer for annotation. All spectra were acquired at each sampling site after adequate haemostasis and were matched with core biopsy samples verified by routine histopathology (the current diagnostic gold standard). After all Raman measurements were complete for each patient the probe tip was cleaned and the wavelength and intensity calibration standards were measured with the fibre optic probe still connected. These data were saved along with the patient data and were used to correct for any changes in the system response between patients.

[0140]

[0114] All biopsies collected as part of the study protocol were submitted and processed at the Department of Histopathology, Imperial College Healthcare NHS Trust, London. Tissue samples underwent standard fixation in formalin and paraffin embedding. 4 pm thick paraffin sections were taken for haematoxylin-eosin staining and immunohistochemistry, and 10 pm thick sections for DNA isolation. Immunohistochemistry was performed according to manufacturer instructions. The following primary antibodies were used: mouse anti-mutant IDH1-R132H (1 :150, DIAH09, Dianova), rabbit anti-ATRX (1 :750, HPA001906, Sigma),

[0141]

[0115] mouse anti-MGMT (1 :500, MAB16200, Chemicon International Inc.). In a subset of cases IDH1 and IDH2 sequencing and / or MS-HRM PCR assay for MGMT promotor methylation were carried out at the Molecular Neuropathology Laboratory of the Division of Neuropathology, Institute of Neurology Queen Square, University College of London.

[0142]

[0116] Data Pretreatment, Identifying Molecular Changes and Multivariate Statistical Analysis The spectral data were pre-treated for standardization before analyses in the following way. First the saved spectral Raman shifts and intensity values were imported into a spreadsheet along with the calibration standard measurements obtained for each patient. The latter were compared to the preset system calibration values and if necessary adjustments were made to that patient’s data to account for any slight differences. A more advanced iterative 5 point quadratic Savitzky-Golay28smoothing was then applied to remove high frequency spectral noise from the calibrated raw data. This was the highest level of smoothing that did not affect the intensity of the narrowest Raman peak (the phenylalanine peak at 1003cm-1) observed in some brain tissue spectra. The corrected raw spectral signal contained the Raman signal superpositioned with a tissue autofluorescence background signal (Figure 4, dark blue line).

[0143]

[0117] An iterative polynomial background subtraction algorithm27was then used to remove this background autofluorescence (Figure 4, red and green dashed lines). The Raman spectrum for each data site was then normalised using the area under the curve (AUG).

[0144]

[0118] For tentative identification of molecular changes the Raman spectra obtained after data the pre-treatment process were averaged by IDH, ATRX and MGMT groups and compared with reference spectra from biomolecular standards. Two main standards were used. The first was a 2-hydroxyglutarate (2-HG) a metabolite which is known to accumulate in IDH mutated cells10. A pure sample of 2-HG was obtained from Sigma-Aldrich Canada Co. (Part # 90790, Oakville, Ontario, Canada) and measured with a sister Raman system to the one described here. The second reference sample was human DNA. This sample was also obtained from was Sigma- Aldrich Canada Co. (Part # D4642) but was measured with a different system29. That system was accurately calibrated allowing for an accurate comparison of relative peak intensities and positions with those measured with the current system.

[0145]

[0119] For multivariate statistical analyses the pretreated data for all Raman spectra were imported into Matlab (MathWorks, MA, USA), and Statistica (TIBCO, CA, USA). A range of supervised and non-supervised machine learning techniques were used for model development including Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA), Random Forrest (RF), Boosted Trees (BT) and Support Vector Machines (SVM) as well as Partial Least Squares / Linear Discriminant Analysis (PLS-LDA). Leave one out cross validation (LOOCV) was used to assess model performance by creating confusion matrices and receiver operating characteristic (ROC) curves. Most of the multivariate analysis methods that were tried gave similar results, although PLS-LDA consistently produced a slightly better prediction and are described here.

[0146]

[0120] The PLS-LDA analytical methods were implemented using MATLAB and have been previously described in detail3031. Briefly, the PLS-LDA-LOOCV procedure consisted of training and testing steps. The LOOCV method builds a classification model on n-1 training spectra, where the remaining spectrum is used as the test spectrum. This procedure is repeated n times such that each spectrum is left out once. The training process included three procedures: (1) the training spectra were standardized by subtracting the training spectra mean and dividing by the training spectra standard deviation, (2) through the NIPALS Algorithm32, the weight factors, loadings, regression coefficient and factor scores of the training spectra were calculated, and (3) a linear discrimination model was developed to predict the classification of new cases. The testing process also included three procedures: (1) the test spectrum was standardised by subtracting the training set mean and by dividing the training set standard deviation, (2) the factor scores of the test spectrum were calculated, and (3) a posterior probability was calculated based on the linear model developed in the training step. The testing process was repeated n times such that each spectrum was treated as the test spectra once. The ROC curve was calculated from the posterior probabilities derived above and represents the diagnostic performance. With good discrimination between 2 groups, the ROC curve moves toward the left and top boundaries of the graph, whereas poor discrimination yields a curve that approaches the diagonal line function. The ALICs were calculated using the trapezoidal rule33.

[0147]

[0121] Examples of the present disclosure will now be described with reference to the following clauses. It should be understood that the following clauses may comprise any of the features described above as well as in the appended claims.

[0148] Clause 1. An apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the apparatus includes: a light source configured to generate at least one light beam; an optical arrangement configured to form the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement is further configured to filter the Raman backscattered optical signal to generate a filtered optical signal; an optical detector configured to receive the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and a data processing arrangement configured to process the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0149]

[0122] Clause 2. An apparatus of clause 1, wherein the apparatus is configured to measure whether or not the region of biological material includes at least one human glioma.

[0150]

[0123] Clause 3. An apparatus of clauses 1 or 2, wherein the apparatus is configured to compute the state of mutation in respect of a measurement of at least one of: an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation.

[0151]

[0124] Clause 4. An apparatus of clause 4, wherein the data processing arrangement is configured to detect the state of mutation as a function of the measurement exceeding a threshold, wherein the threshold is substantially 94%.

[0152]

[0125] Clause 5. An apparatus of any one of the preceding clauses, wherein the apparatus is configured to provide real-time in vivo computation of the state of mutation with a refresh time of less than 10 seconds, optionally less than 2 seconds.

[0153]

[0126] Clause 6. An apparatus of any one of the preceding clauses, wherein the apparatus is configured to compute the state of mutation from a Raman spectrum in a range of 300 to 2000 cm-1.

[0154]

[0127] Clause 7. An apparatus of any one of the preceding clauses, wherein the apparatus is configured for at least one of the optical arrangement and one or more algorithms executed by the data processing arrangement to reduce an influence of extraneous ambient light radiation when computing the state of mutation.

[0155]

[0128] Clause 8. An apparatus of clause 7, wherein the optical arrangement includes one or more optical filters for reducing the influence of extraneous ambient light radiation.

[0156]

[0129] Clause 9. An apparatus of clause 7 or 8, wherein the apparatus is configured to process raw data of each Raman spectrum to apply thereto a background correction, a calibration and smoothing, a removal of autofluorescence, to generate corresponding filtered spectra for the data processing arrangement to process to compute the state of mutation.

[0157]

[0130] Clause 10. An apparatus of any one of the preceding clauses, wherein the data processing arrangement is configured to use a classification model using at least one of partial least squares computation and linear discriminant analysis (PLS-LDA) to compute the state of mutation y way of predicting genetic and epigenetic changes in the region of biological tissue.

[0158]

[0131] Clause 11. An apparatus of any one of the preceding clauses, wherein the optical arrangement is configured to include a flexible probe for a user of the apparatus to use to characterize the region of biological issue, wherein the flexible probe includes a flexible waveguide through which the optical probe beam is guided to the region of biological issue and though which the Raman backscattered optical signal obtained from the region of biological tissue is guided to be processed within the apparatus to compute the state of mutation of the biological tissue, wherein the flexible probed is optionally implemented as a flexible bifurcated optical fibre probe.

[0159]

[0132] Clause 12. An apparatus of any one of the preceding clauses, wherein the optical arrangement and the data processing arrangement are configured to measure a magnitude of at least one of following molecular species: glycogen and cholesterol, nucleic acids and amino acids, lipids (CH3) amides, amide 1.

[0160]

[0133] Clause 13. An apparatus of any one of the preceding clauses wherein the optical detector is implemented ss a CCD, and the light source is implemented as a wavelength-stabilized laser operating to provide substantially 785 nm wavelength output light.

[0161]

[0134] Clause 14. Use of an apparatus of any one of the preceding clauses for determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0162]

[0135] Clause 15. A method for using an apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the method includes: using a light source to generate a light beam; using an optical arrangement to form the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement is configured to filter the Raman backscattered optical signal to generate a filtered optical signal; using an optical detector to receive the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and using a data processing arrangement to process the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

[0163]

[0136] Clause 16. A software product recorded on a machine-readable data storage medium, wherein the software product, when executed on computing hardware, is configured to execute a method of clause 15.

[0164] References

[0165] 1. Rouse, C., Gittleman, H., Ostrom, Q. T., Kruchko, C. & Barnholtz-Sloan, J. S. Years of potential life lost for brain and CNS tumors relative to other cancers in adults in the United States, 2010. Neuro-Oncology 18, 70-77, doi:10.1093 / neuonc / nov249 (2016).

[0166] 2. Deorah, S., Lynch, C. F., Sibenaller, Z. A. & Ryken, T. C. Trends in brain cancer incidence and survival in the United States: Surveillance, Epidemiology, and End Results Program, 1973 to 2001. Neurosurgical Focus 20, E1 ,doi:10.3171 / foc.2006.20.4.E1 (2006).

[0167] 3. Verhaak, R. G. W. et al. An integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, I DH1 , EGFR and NF1. Cancer cell 17, 98, doi: 10.1016 / j.ccr.2009.12.020 (2010).

[0168] 4. DN, L., H, O., OD, W. & WK, C. WHO Classification of Tumours of the Central Nervous System. (2016).

[0169] 5. McLendon, R. et al. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068, doi:10.1038 / nature07385 (2008).

[0170] 6. Disease, N. R. C. U. C. o. A. F. f. D. a. N. T. o. Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease. (National Academies Press (US), 2011).

[0171] 7. Drew, L. Pharmacogenetics: The right drug for you. Nature 537, S60, doi:10.1038 / 537S60a (2016).

[0172] 8. Lui, H., Zhao, J., McLean, D. & Zeng, H. Real-time Raman Spectroscopy for In Vivo Skin Cancer Diagnosis. Cancer Research 72, 2491-2500, doi: 10.1158 / 0008- 5472. CAN- 11-4061 (2012).

[0173] 9. Santos, I. P. et al. Raman spectroscopy for cancer detection and cancer surgery guidance: translation to the clinics. The Analyst 142, 3025-3047, doi:10.1039 / C7AN00957G (2017).

[0174] 10. Dang, L. et al. Cancer-associated IDH1 mutations produce 2- hydroxyglutarate. Nature 462, 739, doi:10.1038 / nature08617 (2009).

[0175] 11. Kong, K., Kendall, C., Stone, N. & Notingher, I. Raman spectroscopy for medical diagnostics — From in-vitro biofluid assays to in-vivo cancer detection. Advanced Drug Delivery Reviews 89, 121-134, doi:10.1016 / j.addr.2015.03.009 (2015).

[0176] 12. Lewis, P. W., Elsaesser, S. J., Noh, K. M., Stadler, S. C. & Allis, C. D. Daxx is an H3.3-specific histone chaperone and cooperates with ATRX in replicationindependent chromatin assembly at telomeres. Proc Natl Acad Sci U S A 107, 14075- 14080, doi: 10.1073 / pnas.1008850107 (2010).

[0177] 13. Kannan, K. et al. Whole exome sequencing identifies ATRX mutation as a key molecular determinant in lower-grade glioma. Oncotarget 3, 1194-1203 (2012).

[0178] 14. Thon, N., Kreth, S. & Kreth, F.-W. Personalized treatment strategies in glioblastoma: MGMT promoter methylation status. OncoTargets and therapy 6, 1363- 1372, doi:10.2147 / GTT.S50208 (2013).

[0179] 15. Jermyn, M. et al. Intraoperative brain cancer detection with Raman spectroscopy in humans. 7, 1-10 (2015).

[0180] 16. Gajjar, K. et al. Diagnostic segregation of human brain tumours using Fourier- transform infrared and / or Raman spectroscopy coupled with discriminant analysis. Analytical methods advancing methods and applications 5, 89-102, doi:10.1039 / C2AY25544H (2012).

[0181] 17. Vogelbaum, M. A. et al. Application of Novel Response / Progression Measures for Surgically Delivered Therapies for GliomasResponse Assessment in NeuroOncology (RANO) Working Group. Neurosurgery 70, 234-244, doi:10.1227 / NEU.0b013e318223f5a7 (2012).

[0182] 18. McCulloch, P. et al. No surgical innovation without evaluation: the IDEAL recommendations. The Lancet 374, 1105-1112, doi:10.1016 / S0140-6736(09)61116- 8 (2009).

[0183] 19. Huang, Z., Zeng, H., Hamzavi, I., McLean, D. I. & Lui, H. Rapid near-infrared Raman spectroscopy system for real-time in vivo skin measurements. Optics Letters 26, 1782-1784, doi:10.1364 / OL.26.001782 (2001).

[0184] 20. Short, M. A. et al. Development and preliminary results of an endoscopic Raman probe for potential in vivo diagnosis of lung cancers. Optics Letters 33, 711- 713, doi:10.1364 / OL.33.000711 (2008).

[0185] 21 . Short, M. A., Wang, W., Tai, I. T. & Zeng, H. Development and in vivo testing of a high frequency endoscopic Raman spectroscopy system for potential applications in the detection of early colonic neoplasia. Journal of Biophotonics 9, 44-48, doi:10.1002 / jbio.201500205 (2016).

[0186] 22. Guze, K. et al. Parameters defining the potential applicability of Raman spectroscopy as a diagnostic tool for oral disease. Journal of Biomedical Optics 14, 014016, doi:10.1117 / 1.3076195 (2009).

[0187] 23. Zhao, J. et al. Clinical Raman measurements under special ambient lighting illumination. Journal of Biomedical Optics 19, 111609-111609, doi:10.1117 / 1.JBO.19.11.111609 (2014).

[0188] 24. ANSI Z136.1 (2014) - Safe Use of Lasers (Electronic Version), 2015).

[0189] 25. NIST: Basic Atomic Spectroscopic Data - Element Names.

[0190] 26. Choquette, S. J., Etz, E. S., Hurst, W. S., Blackburn, D. H. & Leigh, S. D. Relative intensity correction of Raman spectrometers: NIST SRMs 2241 through 2243 for 785 nm, 532 nm, and 488 nm / 514.5 nm excitation. Applied spectroscopy 61 , 117— 129 (2007).

[0191] 27. Zhao, J., Lui, H., McLean, D. I. & Zeng, H. Automated Autofluorescence Background Subtraction Algorithm for Biomedical Raman Spectroscopy. Applied Spectroscopy, doi: 10.1366 / 000370207782597003 (2016).

[0192] 28. Savitzky, A. & Golay, M. J. E. Smoothing and Differentiation of Data by Simplified Least Squares Procedures. Analytical Chemistry 36, 1627-1639, doi:10.1021 / ac60214a047 (1964).

[0193] 29. Short, M. A. et al. Changes in nuclei and peritumoral collagen within nodular basal cell carcinomas via confocal micro-Raman spectroscopy. J Biomed Opt 11 , 34004, doi: 10.1117 / 1.2209549 (2006).

[0194] 30. McGregor, H. C. et al. Real-time endoscopic Raman spectroscopy for in vivo early lung cancer detection. Journal of Biophotonics 10, 98-110, doi: 10.1002 / jbio.201500204 (2017).

[0195] 31. Zhao, J., Zeng, H., Kalia, S. & Lui, H. Wavenumber selection based analysis in Raman spectroscopy improves skin cancer diagnostic specificity. Analyst 141 , 1034- 1043, doi:10.1039 / C5AN02073E (2016).

[0196] 32. Hill, T. & Lewicki, P. Statistics: Methods and Applications. 1 edn, (StatSoft Inc, 2005).

[0197] 33. Hanley, J. A. & McNeil, B. J. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143, 29-36, doi: 10.1148 / radiology.143.1 .7063747 (1982).

Claims

- 32Claims1. An apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the apparatus comprises: a light source configured to generate at least one light beam; an optical arrangement configured to form the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement is further configured to filter the Raman backscattered optical signal to generate a filtered optical signal; an optical detector configured to receive the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and a data processing arrangement configured to process the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

2. An apparatus of claim 1, wherein the apparatus is configured to determine whether or not the region of biological tissue includes at least one human glioma, tumour and / or cancerous cells based on the state of mutation of the biological tissue.

3. An apparatus of claim 1 or 2, wherein the apparatus is configured to compute the state of mutation in relation to at least one of: an IDH mutation, an occurrence of ATRX loss, an occurrence of MGMT methylation, based on the detector signal.

4. An apparatus of any one of the preceding claims, wherein the apparatus is configured to provide real-time in vivo computation of the state of mutation with a refresh time of less than 10 seconds, optionally less than 2 seconds, or optionally less than 1 second.

5. An apparatus of any one of the preceding claims, wherein the apparatus is configured to compute the state of mutation from a Raman spectrum in a range of 300 to 2000 cm-1.- 336. An apparatus of any one of the preceding claims, wherein at least one of the optical arrangement and one or more algorithms when executed by the data processing arrangement are configured to reduce an influence of extraneous ambient light radiation when computing the state of mutation, optionally wherein the optical arrangement is configured to reduce the influence of extraneous ambient light radiation by comprising one or more optical filters.

7. An apparatus of claim 6, wherein the apparatus is configured to apply to the detector signal a background correction, a calibration and smoothing, a removal of autofluorescence, to generate corresponding filtered spectrum or spectra for the data processing arrangement to process to compute the state of mutation.

8. An apparatus of any one of the preceding claims, wherein the data processing arrangement is configured to execute a classification model using at least one of partial least squares computation and linear discriminant analysis, PLS-LDA, to compute the state of mutation in the region of biological tissue.

9. An apparatus of any one of the preceding claims, wherein the optical arrangement is configured to include a flexible probe for a user of the apparatus to use to characterize the region of biological tissue, wherein the flexible probe includes a flexible waveguide through which the optical probe beam is guided to the region of biological issue and through which the Raman backscattered optical signal obtained from the region of biological tissue is guided to be processed within the apparatus to compute the state of mutation of the biological tissue, wherein the flexible probed is optionally implemented as a flexible bifurcated optical fiber probe.

10. An apparatus of any one of the preceding claims, wherein the data processing arrangement is configured to determine a concentration of at least one of following molecular species: glycogen, cholesterol, nucleic acids, amino acids, lipids, (CH2), (CH3), amides, and amide 1.

11. An apparatus of any one of the preceding claims wherein the optical detector is implemented as a charge-coupled device, CCD, and / or the light source isimplemented as a wavelength-stabilized laser operating to provide a light beam of 785 nm wavelength.

12. Use of an apparatus of any one of the preceding claims for determining from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

13. A method for an apparatus for in vivo profiling a region of biological tissue to determine its state of mutation, wherein the apparatus comprises a light source, optical arrangement, optical detector and a data processing arrangement, the method includes: the light source generating a light beam; the optical arrangement forming the light beam into an optical probe beam for illuminating the region of biological tissue and receiving a Raman backscattered optical signal from the region of biological tissue, wherein the optical arrangement filtering the Raman backscattered optical signal to generate a filtered optical signal; the optical detector receiving the filtered optical signal to generate a corresponding detector signal including one or more components indicative of one or more molecular species present in the region of biological tissue; and the data processing arrangement processing the detector signal to determine from the one or more components a concentration of the one or more molecular species to compute the state of mutation of the biological tissue.

14. A computer-readable medium having instructions stored thereon which when executed by one or more processors, cause the one or more processors to implement a method according to claim 13.

Images