Morphometric genotyping of cells using optical tomography to detect tumor mutational burden

Abstract

The present invention discloses a method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB). This method provides a non-invasive method for characterizing TMB, which is responsive to tumors in the early stages of tumor development and is independent of tumor size. This method allows cancer treatment to be tailored to the specific characteristics of the cancer a patient may have, resulting in more effective cancer management with far fewer side effects.

Description

Technical Field

[0001] The present invention relates to optical tomography at the cellular and subcellular scales. More specifically, the present invention relates to a system and method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB). Background Art

[0002] Alterations in nuclear morphology have been the primary histopathological biomarker for cancer detection for the past 140 years. Numerous published studies have demonstrated a direct link between chromatin organization in the cell nucleus and cellular function at the levels of DNA replication, translation, and protein expression. 1-3 In particular, chromatin organization, which underlies the nuclear 3D architecture, is considered a major factor influencing regional and global mutation rates in human cancer cells. 4-5 .

[0003] Current treatments for many cancers involve targeting immune system checkpoint inhibitors called CTLA4, PD-1, and PD-L1, proteins involved in enabling tumors to evade immune system responses. 6 Although durable responses have been achieved with immunotherapy for many solid tumors, only a subset of patients actually benefit. For example, the following are the response rates to single-agent PD-1 / PD-L1 inhibition: 40% for melanoma 7、8 , non-small cell lung cancer (NSCLC) is 25% 9、10 , renal cell carcinoma 19% 11 Furthermore, current immunotherapies carry a strong risk of adverse side effects. 12-14 Therefore, there is a need for reliable biomarkers that can reliably predict which patients will benefit from immunotherapy to reduce the unnecessary burden of inflammation and immune-related adverse reactions on patients. Several biomarkers have been identified that help predict patient response to immunotherapy. 15 One of these biomarkers is PD-L1 expression, which is necessary for treatment response but insufficient to determine response due to tumor heterogeneity and measurement of expression levels. 16-21 Recently, mismatch repair (MMR) deficiency and tumor mutational burden (TMB; the number of somatic, coding, base substitution, and indel mutations per megabase of genomic DNA) have been found to be good predictors of response to immunotherapy. 15、22-27 MMR deficiency leads to genomic and microsatellite instability (MSI) and high TMB, resulting in the expression of neoantigens, which make tumor cells more susceptible to attack by cytotoxic T cells. 23 Challenges in detecting TMB from solid tumors or ctDNA include the inability to perform biopsies, sensitivity for detecting early-stage disease, and lack of consistency in data obtained using paired tissues and different NGS (next-generation sequencing) platforms.28、29 As described below, a more rapid, minimally invasive, and inexpensive approach uses VisionGate Cell-CT technology to detect low versus high TMB in cancer cells based on the potential of TMB to confer morphometric alterations to structural biomarkers, which can be quantified at the submicron spatial scale using 3D light microscopy. The demonstrated link between chromatin organization and genomic DNA mutation rates suggests the presence of structural biomarkers in the cell nucleus that could be used to detect genomic instability and / or more specific types of genomic alterations in cancer.

[0004] The ability to detect MMR deficiency and TMB levels by measuring structural biomarkers is supported by numerous studies reporting histopathological differences in colorectal cancers harboring microsatellite mutations due to MMR deficiency. Typically, these tumors are mucinous and poorly differentiated, composed of relatively large, round, and regular cells with abundant amphipathic cytoplasm. 30-32 In addition, Alexander et al. 33 reported that colon cancers with MSI exhibit signet ring cells and cribriform formations. Gisselsson et al. 34 A study by researchers from the University of California, Berkeley, found that abnormal nuclear shape is an indicator of genetic instability in short-term tumor cell cultures. In cultures of 58 tumors from bone, soft tissue, and epithelial cells, nuclear blebs, strands, and micronuclei occurred more frequently in tumors harboring genetic instability factors. Other cell culture systems that have revealed a link between DNA repair and nuclear morphology include the following studies. Bai et al. 35 reported that treatment of HeLa cells in which Rad9 expression had been knocked out with alkylating agents resulted in abnormal nuclear morphology. Debes et al. 36 showed that transfection of p300 into prostate cancer cells in culture caused quantifiable nuclear changes, such as diameter, circumference, and absorbance. The p300 gene and the highly homologous CREB binding protein (CBP) gene are mutated together in >85% of microsatellite instability (MSI)+ colon cancer cell lines. 37 , and loss of heterozygosity at the p300 locus has been observed in advanced intestinal-type gastric cancer. 38

[0005] Another structural biomarker associated with tumor progression is the centrosome. MMR deficiency and genomic instability are closely associated with an increase in the number of structurally abnormal centrosomes. 39、40 Centrosomes play crucial roles in many microtubule-mediated processes, such as determining cell shape and polarity. 41-44 .

[0006] As mentioned above, morphometric alterations based on defects in MMR and MSI have been implicated in various cancer types. Although the data presented below establish the Cell-CT TM The platform's ability to morphometrically genotype lung adenocarcinoma cell lines harboring diverse driver mutations was demonstrated, but the utility of this technology should allow for the identification of MMR defects and mutational burden in a variety of cancers.

[0007] In a related development, Nelson has employed optical tomography techniques for 3D imaging of biological cells, as disclosed, for example, in U.S. Patent No. 6,522,775, entitled “Apparatus and Method for Imaging Small Objects in a Flow Stream Using Optical Tomography,” issued February 18, 2003, the entire disclosure of which is incorporated herein by reference. Further significant developments are taught in U.S. Patent No. 7,738,945, entitled “Method and Apparatus for Pseudo-Projection Formation for Optical Tomography,” issued June 15, 2010 (Fauver '945), and U.S. Patent No. 7,907,765, entitled “Focal Plane Tracking for Optical Microtomography,” issued March 15, 2011 (Fauver '765), the entire disclosures of which are incorporated herein by reference. The early lung cancer detection technology has been fully developed and commercialized by VisionGate, Inc. of Phoenix, Arizona, to provide measurement advantages that have demonstrated significant improvements in the operating characteristics of conventional morphological cytology analysis.

[0008] Processing in such an optical tomography system begins with sample collection and preparation. For diagnostic applications in lung disease, patient samples can be collected non-invasively in a clinic or at home. In the clinical laboratory, the sample is processed to remove non-diagnostic material, fixed and then stained. The stained sample is then mixed with an optical gel and the suspension is injected into a microcapillary tube. Images of objects (e.g., cells) in the sample are collected as the cell is rotated 360 degrees relative to the image acquisition optics in the optical tomography system. The resulting image includes a set of extended depth of field images from different viewing angles, called "pseudo-projection images." The set of pseudo-projection images can be mathematically reconstructed using back-projection and filtering techniques to produce a 3D reconstruction of the target cell. Having equidistant or approximately equal resolution in all three dimensions is an advantage of 3D tomographic cell imaging, particularly for quantitative feature measurement and image analysis. Building on the teachings therein, VisionGate, Inc. of Phoenix, Arizona, has developed an early lung cancer detection technology to provide measurement advantages that have the potential to greatly improve the operating characteristics of conventional morphological cytological analysis. Published clinical data 45、46 Display, using Cell-CT TM Non-invasive sputum analysis performed by the platform can detect early-stage lung cancer with high sensitivity (92%) and specificity (95%).

[0009] The 3D reconstructed digital image can then be analyzed for quantification by measuring subcellular structures, molecules, or molecular probes of interest. Objects such as biological cells can be stained or labeled with at least one absorptive contrast agent or labeled molecular probe, and the measured amount and structure of the biomarker can yield important information about the disease state of the cell, including but not limited to various cancers, such as lung, breast, prostate, cervical, gastric, and pancreatic cancers, as well as various stages of dysplasia.

[0010] However, until the disclosure herein, there has been no reliable method for detecting TMB using optical tomography. By providing methods and systems herein for identifying TMB in target cells, patients may benefit from treatment with immunomodulators (e.g., iloprost) to reduce their risk of developing lung cancer. Summary of the Invention

[0011] This overview is provided to introduce some concepts in a simplified form, which are further described in the detailed description below. This overview is not intended to identify key features of the claimed subject matter, nor is it intended to be used to help determine the scope of the claimed subject matter. The invention presented in this disclosure describes a method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB). TMB has been found to be important for triaging cancer patients for appropriate cancer treatment. The method provides a non-invasive method for characterizing TMB, which is responsive to tumors in the early stages of tumor development and is independent of the size of the tumor. Therefore, the present invention has important implications for the development of cancer treatments tailored to specific characterizations of the cancer a patient may have, thereby allowing for more effective cancer management with fewer side effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] While the novel features of the present invention are particularly set forth in the appended claims, the organization and content of the present invention, as well as other objects and features, will be better understood and appreciated from the following detailed description when taken in conjunction with the accompanying drawings, in which:

[0013] Figure 1 A functional overview of the lung cancer test used for sample analysis is shown schematically.

[0014] Figure 2 The basic system components of a 3D optical tomography system used in a lung cancer testing system are shown schematically.

[0015] Figures 3A to 3C A single perspective view of a 3D image of an adenocarcinoma cell is shown.

[0016] Figure 4 Cilia on lung columnar cells are shown.

[0017] Figure 5 ROC curves for sensitivity versus specificity of the abnormal cell classifier are shown.

[0018] Figure 6 An example of a classification cascade used to identify specific mutations associated with different cancer types is shown schematically.

[0019] Figure 7 The results of the experimental study are listed in the table, which show the area under the ROC (aROC) as well as the sensitivity and specificity for the target cells.

[0020] Figure 8 A flow chart illustrating an example of a method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB) is schematically shown.

[0021] Figure 9Schematically shown is a flow chart of an example of a method for developing one or more morphometric classifiers to use tumor mutation burden (TMB) as ground truth.

[0022] Figure 10 A flow chart schematically illustrates an example of a method for treating a malignancy in a human subject using immunotherapy.

[0023] In the drawings, like reference numerals denote similar elements or components. The sizes and relative positions of elements in the drawings are not necessarily drawn to scale. For example, the shapes and angles of various elements are not drawn to scale, and some elements are arbitrarily enlarged and positioned to improve the readability of the drawings. Furthermore, the particular shapes of the elements depicted are not necessarily intended to convey any information about the actual shape of the particular element and are selected solely for ease of identification in the drawings. DETAILED DESCRIPTION

[0024] The following disclosure describes a method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB). Several features of the methods and systems according to exemplary embodiments are illustrated and described in the accompanying drawings. It will be understood that the methods and systems according to other exemplary embodiments may include additional processes or features different from those shown in the drawings. Exemplary embodiments are described herein with respect to optical tomography cell imaging systems. However, it should be understood that these examples are for illustrative purposes only and the present invention is not limited thereto.

[0025] The present invention provides an early lung cancer detection system that measures TMB using samples processed by an optical tomography system. This system produces equidistant, submicron-resolution 3D cell images, which are then processed by automated feature extraction and classification algorithms to identify abnormal cells with high accuracy. Because abnormal cells are rare and non-diagnostic contaminants are numerous, only a system capable of performing cell-based detection with high sensitivity and very high specificity can effectively manage lung cancer detection while ensuring sufficient sample availability.

[0026] definition

[0027] Generally, as used herein, the following terms have the following meanings, unless the context dictates otherwise:

[0028] When used in conjunction with the term "comprising" in the claims or the specification, the use of the words "a" or "an" means one or more than one, unless the context dictates otherwise. The term "about" refers to the stated value plus or minus the range of measurement error, or, if no measurement method is specified, plus or minus 10%. The use of the term "or" in the claims is used to mean "and / or" unless it is expressly stated that only alternatives are referred to or the alternatives are mutually exclusive. The terms "include," "have," "include," and "contain" (and variations thereof) are open-ended linking verbs and allow the addition of other elements when used in the claims.

[0029] Throughout this specification, references to "one example," "an exemplary embodiment," "one embodiment," "an example," or combinations and / or variations of these terms mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the present disclosure. Thus, appearances of the phrases "in one embodiment" or "in an example" throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0030] “Adequacy” refers to the content of the sample and defines the limit of target cells to determine whether enough cell pellets have been analyzed.

[0031] As used herein, "calcitriol" is the synthetic (man-made), active form of vitamin D3 (cholesterolcalciferol).

[0032] "Capillary" has its generally accepted meaning and is intended to include transparent microcapillaries and equivalents having an inner diameter of typically 500 microns or less, although larger diameters may also be used.

[0033] "Cell" refers to a biological cell, such as a human, mammalian or animal cell.

[0034] “Cell-CT TM "Platform" refers to an optical tomography system manufactured by VisionGate, Inc. of Phoenix, Arizona, which incorporates the teachings of the Nelson and Fauver patents referenced herein above and improvements to those teachings. Cell-CT TM The platform is an automated high-resolution 3D tomography microscope and computing system for imaging flow cells. Compared with traditional optical imaging methods, Cell-CT TMThe platform computes 3D cell images with equal spatial resolution in all dimensions (isotropic resolution), making measurements independent of orientation. Furthermore, it eliminates the focal plane ambiguity and view orientation dependence typical of conventional microscopy, providing information content that automatically identifies various cell types and clearly identifies rare abnormal cells within a predominantly normal cell population.

[0035] "CellGazer" refers to a software-based utility that facilitates the viewing of 2D and 3D images of cells provided by Cell-CT. The result of the cytometry is a detailed differential diagnosis of cell types, which is then followed by LuCED testing to determine the final outcome of the case.

[0036] As used herein, "chimeric antigen receptor (CAR)" refers to an artificial T cell receptor (also called a chimeric T cell receptor, or chimeric immunoreceptor) that is an engineered receptor that grafts arbitrary specificity onto an immune effector cell.

[0037] As used herein, "CIS" has its generally accepted meaning of carcinoma in situ, also known as tumor in situ.

[0038] "Depth of field" is the length along the optical axis over which the focal plane may move before producing unacceptable image blur for a given feature.

[0039] "Enrichment" refers to the process of extracting target cells from the original sample. This process creates an enriched pellet whose cells can then be more efficiently imaged on the Cell-CT system.

[0040] As used herein, "immunotherapy" is applicable to the field of oncology and refers to a method for improving, treating or preventing malignant tumors in human subjects, wherein the effect of the method assists or enhances the immune system in eradicating cancer cells, including the administration of cells, antibodies, proteins or nucleic acids that activate an active (or achieve a passive) immune response to destroy cancer cells. It also includes the combination treatment of biological adjuvants (such as interleukins, cytokines, Bacillus Calmette-Guerin, monophosphoryl lipid A, etc.) with conventional therapies for treating cancer, such as chemotherapy, radiotherapy or surgery, which prevent or destroy the growth of cancer cells by activating the immune system, as well as in vivo, ex vivo and adoptive immunotherapy (including the use of autologous and / or allogeneic cells or immortalized cell lines).

[0041] As used herein, "iloprost" is an immunomodulatory agent that comprises a synthetic analog of prostacyclin PGI2.

[0042] “ Test" refers to the use of Cell-CTTM An early lung cancer detection test is based on a platform developed by VisionGate, Inc. of Phoenix, Arizona, which incorporates the teachings of the Nelson and Fauver patents mentioned above and improvements upon those teachings.

[0043] “ The “process” refers to the mechanism of 3D cell reconstruction, classification of abnormal cells, and pathological confirmation.

[0044] "LDCT" refers to a low-dose computed tomography (CT) radiographic scan.

[0045] "Subject" means a single cell, human cell, mammalian cell, article, thing, or other entity.

[0046] A "pseudo-projection" comprises a single image representing a sampled volume having an extent greater than the native depth of field of the optical system, wherein the pseudo-projected image thus formed comprises the integral of the image extent from a fixed viewpoint focal plane. The concept of pseudo-projection is taught in Fauver '945.

[0047] "Sample" means the entire product obtained from a single test or procedure on a single patient (e.g., sputum, biopsy, or nasal swab submitted for analysis). A sample may consist of one or more objects. The results of the diagnosis of the sample become part of the diagnosis of the case.

[0048] "ROC" has its generally accepted meaning of receiver operating characteristic.

[0049] "Specimen" refers to an intact cell preparation ready for analysis, including whole or partial aliquots or samples.

[0050] As used herein, "subject" refers to a human patient.

[0051] "Target cells" are cells from a sample whose characterization or enumeration is specifically desired. For example, in the LuCED test, the target cells are normal bronchial epithelial cells. A minimum number of samples must be enumerated during the test for the sample to be considered adequate.

[0052] As used in the context of image processing, a "threshold" includes a decision boundary value for any measurable characteristic of a feature. The threshold can be predetermined or set based on instrument specifications, acceptable error rates, statistical data, or other criteria based on recognized pattern recognition principles.

[0053] Tumor mutation burden (TMB) refers to the number of somatic, coding, base substitution, and indel mutations per megabase pair of genomic DNA.

[0054] "TNM stage" is used in the context of lung cancer with its generally accepted meaning and refers to the tumor, node, metastasis (TNM) stage as defined by medical societies such as The International Association for the Study of Lung Cancer (IASLC).

[0055] Vorinostat, also known as suranilide aminohydroxamic acid, is commonly used as a histone deacetylase (HDAC) inhibitor in Barrett's esophagus.

[0056] As used in the context of image processing, a "voxel" is a volume element on a 3D grid.

[0057] Overview

[0058] refer to Figure 1 , schematically shows a functional overview of a pulmonary dysplasia and cancer test system for sample analysis. The test system 5 includes an apparatus and method 10 for sample collection, followed by a test 12 for early lung cancer detection, such as The early lung cancer test 12 also includes equipment and methods for sample staining and enrichment 14, 3D cell imaging 20, 3D cell sorting 22, and clinician review 25 of abnormal candidate cells.

[0059] If sputum is used, it is usually collected by spontaneous coughing at home or by induction in the clinic. Other types of sample collection, such as biopsies, can be performed in clinical settings. The specimen is processed to remove contaminants and non-bronchial epithelial cells, such as by reducing the size of white blood cells and oral squamous cells. The enriched sample is then analyzed in Cell-CT. TM The processing is performed on a platform that digitally images cells in true 3D with isometric, submicron resolution, as described by Nelson and Fauver above. Cancer-related biosignatures are measured on the 3D cell images and combined into a score that identifies a small number of cells with cancerous characteristics. These cells are then optionally displayed for analysis using an inspection station such as the CellGazer developed by VisionGate, Inc. of Phoenix, Arizona. TM The checkpoint provides a visual display that allows the cytologist to view cell images in 2D and 3D to establish a definitive normal or abnormal state for a particular cell candidate. Three-dimensional (3D) cell classification can be performed using the technology disclosed below.

[0060] The cell imaging system 20 includes processes implemented by computer software, such as executed by a personal computer interfaced with the optomechanical device, to correct for motion that occurs during image capture. Most cell images emerge from the filtered backprojections as well-reconstructed images. A computer-implemented algorithm can identify poorly reconstructed cells so that they can be removed from subsequent processing. An example of such a method for detecting poor reconstructions is taught in U.S. Patent No. 8,155,420, issued April 10, 2012, to Meyer et al., entitled "System and Method for Detecting Poor Quality in 3D Reconstructions," the disclosure of which is incorporated herein by reference.

[0061] Early attempts to develop lung cancer screening programs were based on sputum cytology, which showed that the sensitivity of the human eye for detecting the disease was insufficient (average, about 60%), but the specificity was very good (Schreiber and McCrory (2003) Chest 123(1 Suppl): 115). This experience led some to conclude that sputum was of no value for detecting lung cancer. Careful analysis involving paraffin-embedded sputum (Booking A, Biesterfeld S, Chatelain R, Gien-Gerlach G, Esser E., Diagnosis of bronchial carcinoma on sections of paraffin-embedded sputum. Sensitivity and specificity of an alternative to routine cytology, Acta Cytol. 1992; 36(1): 37-47) showed that in 86% or more of cancer patients, the samples actually contained abnormal cells. The best results were obtained when the sample was collected during a morning cough on three consecutive days. Further analysis showed that sputum contained abnormal cells stratified by all relevant clinical factors, including tumor histological type, size, stage, and location (Neumann T, Meyer M, Patten F, Johnson F, Erozan Y, Frable J, et al., Premalignant and Malignant Cells in Sputum from Lung Cancer Patients, Cancer Cytopathology, 2009; 117(6): 473-481). Based on these sample characteristics, currently published lung cancer detection tests use spontaneous cough sputum. Preliminary evaluations have shown satisfactory results with sputum fixation using Cytoyt (Hologic, Marlborough, MA) or the well-known Saccomanno method. The issue of sample adequacy is also important for sputum cytology. Attempts to increase the amount of sputum have also had varying success. Sputum induction can increase sputum production to help obtain an adequate overall sample.

[0062] Example of sample enrichment and preparation

[0063] In a lung cancer detection test suitable for TMB detection Show In this example, samples underwent three processing stages before analysis: 1) cell isolation and cryopreservation; 2) enrichment via fluorescence-activated cell sorting (FACS); and 3) embedding the enriched cells in optical oil that was refractive-index matched to the optical components of the optical tomography imaging system.

[0064] Cryopreservation and FACS enrichment (FACS is an example)

[0065] Sputum was treated with the mucolytic agent dithiothreitol (DTT) (Fisher Scientific, Waltham, MA). In one example, for long-term storage, the sample was filtered through a 41 μm nylon mesh and maintained at -80°C in 15% dimethyl sulfoxide (DMSO) (Fisher Scientific, Waltham, MA). After filtration, aliquots of up to 100 μL were taken out for lung cancer detection test analysis. First, sputum cells were stained with hematoxylin (Electron Microscopy Sciences, Hatfield, PA) for downstream lung cancer detection test imaging. The cells were then treated with an antibody mixture containing a fluorescent conjugate selected for enrichment of bronchial epithelial cells and depletion of contaminating inflammatory cells (neutrophils and macrophages). An anti-cytokeratin-FITC conjugate mixture (Cell Signaling, Danvers, MA) targets cytokeratins expressed in normal and malignant epithelial cells. Anti-CD45-APC conjugate (Mylteni, Bergisch-Gladbach, Germany) was used to negatively select for inflammatory cells. Before sorting, cells were also stained with DAPI (Life Technologies, Grand Island, NY). For FACS enrichment, a DAPI-positive mother gate was created to exclude doublets and debris, and then high side scatter events, which were mainly oral squamous cells, were excluded. Subsequently, a cytokeratin high (high FITC) and CD45 low (low APC) sub-gate was drawn. The cell population in this sub-gate is the enriched target epithelial cells, and the cells were analyzed using an optical tomography system (e.g., Cell- Optical tomography system) for sorting to enable more efficient downstream lung cancer detection test analysis.

[0066] Embedding of enriched cells

[0067] After FACS enrichment (or any other enrichment process), the cells are dehydrated in ethanol and then suspended in xylene. The cells are then transferred to a suitable volume of optical medium and embedded therein. The optical medium is a viscous oil with a refractive index that matches the optical tomography system. Once embedded, the cells are injected into a disposable cartridge for imaging on the optical tomography system.

[0068] Now refer to Figure 2, showing the basic system components of a 3D optical tomography imaging system used in a lung cancer testing system. The cell imaging system 20 is an automated, high-resolution 3D tomography microscope and computing system for imaging cells in flow. It includes an illumination source 90 optically coupled to a condenser 92, which optically cooperates with an objective lens 94 to scan an image of an object 1 contained in a capillary tube 96. The image is acquired by scanning the volume occupied by the object with an oscillating mirror 102 and transmitted to a high-speed camera 106 via a beam splitter 104. The high-speed camera generates a plurality of pseudo-projection images 110. For each object, a set of pseudo-projection images is generated for multiple axial tube rotation positions.

[0069] Although the test system is not limited to any one contrast method, in one example, a lung cancer detection test is specific to cell morphology based on the traditionally used hematoxylin stain. In the lung cancer detection test application, the optical tomography system computes a 3D cell image with the same resolution in all dimensions (i.e., isotropic resolution), making the measurement independent of orientation. In addition, the focal plane ambiguity and view orientation dependence typical of conventional microscopes are eliminated, providing information content that automatically identifies various types of cells and clearly identifies rare abnormal cells within a predominantly normal cell population. The output of the optical tomography system identifies approximately 0.5% of all cells as being to be detected using the CellGazer TM The researchers were identified as outliers using a VisionGate (Phoenix, AZ) workstation, an imaging software tool that enables humans to view images free of focal plane and orientation ambiguities.

[0070] Optical tomography systems image small amounts of liquid suspensions. For the lung cancer detection test, these cells were derived from the enriched epithelial cell population described above. Because optical tomography systems can separate closely aligned objects, the narrow focus of a single-file cell flow, as required by standard flow cytometers, is not required.

[0071] Exemplary operation of the lung cancer testing system is described in the Nelson and Fauver references incorporated by reference above, as well as in other patents, including U.S. Patent No. 8,254,023, issued to Watson et al. on August 28, 2012, entitled "Optical Tomography System with High-Speed Scanner," which is also incorporated herein by reference. In operation, stained nuclei of biological cells 1 are suspended in an optical medium 112 and injected into a capillary tube 96 having an inner diameter of, for example, 62 μm. The capillary system is designed to be disposable, thus eliminating the possibility of cross-contamination between samples. Pressure 114 applied to the fluid moves the object 1 into the imaging position, and 3D data is then acquired as the tube rotates. Mirror 102 is actuated to sweep the focal plane across the object, and the image is integrated by a camera to produce pseudo-projections from each individual viewpoint. The glass holder connecting capillary tube 96 to the optical tomography system is not shown. The holder has a hole in the middle with a diameter slightly larger than the outer diameter of the capillary and glass plate (not shown to simplify the diagram) to facilitate optical coupling with the objective lens and condenser. The capillary tube containing the cell embedded in the transmission medium is threaded through the holder. The transmission medium, the glass capillary tube, the capillary holder, the oil that interfaces with the lens, and the lens itself are all made of materials with the same optical index. Therefore, when the capillary tube rotates 360 degrees, while the cell can rotate to allow the capture of a collection of 500 pseudo-projections, the light passes through the optics of the optical tomography system, the capillary tube, and the cell without refraction. Because the cells are suspended in the liquid medium, they are susceptible to a small amount of movement when the pseudo-projection images 110 are acquired.

[0072] Therefore, the cell images in the pseudo-projection must be registered to a common center so that the cell features enhance each other during the reconstruction process. U.S. Patent No. 7,835,561, entitled "Method for Image Processing and Reconstruction of Imagesfor Optical Tomography", discloses an error correction technique for pseudo-projection. U.S. Patent No. 7,835,561 is incorporated herein by reference. The pseudo-projection set after correction is processed using a filtered back-projection algorithm similar to that used in conventional X-ray CT to calculate tomography 3D cell reconstruction. Pseudo-projection images 110 taken at three angular positions (0g, 90g and 180g) are shown. Light source 90 provides illumination at a wavelength of 585nm to optimize image contrast based on the hematoxylin absorption spectrum. In reconstruction, 3D pixels or voxels are cubic, with a size of approximately 70nm in each dimension. The size of the reconstruction volume will vary because the image acquisition volume can be cropped around the object. Typically, the volume on the side is approximately 200 to 300 pixels.

[0073] Now refer to Figures 3A to 3C , showing a perspective view of a 3D image of an adenocarcinoma cell. Figure 3A Adenocarcinoma cells are shown in maximum intensity projection (13) Since grayscale values in 3D images are associated with various cellular features, a lookup table mapping cellular structures to color and opacity values was built to map the cellular structures to the color and opacity values in the center (e.g. Figure 3B as shown) and on the right (as shown Figure 3C Cell images are generated using a color reproduction of these images (as shown). In the color reproduction of these images, translucent white 402 represents the cytoplasm, opaque blue 404 represents the cell nucleus, translucent green 406 represents the loose chromatin and nucleoplasm, and condensed chromatin and nucleoli are represented by opaque red 408. In international patent regulations that provide only black and white drawings, these colors have been identified by borders (shown as dashed borders) identified by the corresponding reference numerals 404, 406, and 408.

[0074] Now refer to Figure 4 , showing cilia on lung columnar cells. The imaged normal bronchial epithelial cells have a single ciliary strand with a diameter of approximately 250 nm. This further demonstrates the resolution of the 3D cell imaging system.

[0075] Now refer to Figure 5 , shows the ROC curve for an abnormal cell classifier. The ROC curve 700 is a plot of sensitivity for dysplastic cells on the vertical axis 701 versus specificity on the horizontal axis 703. Point 707 represents the area where the dysplastic cell classifier performs at 75% sensitivity at a specificity close to 100%. The classifier was constructed using a data set containing cells indicative of abnormal lung processes, including moderate to severe dysplasia and certain atypical cell conditions. The classifier was trained using a set of approximately 150 known dysplastic cells and approximately 25,000 known normal cells. The accuracy is demonstrated by the single cell ROC curve 700, which shows near-perfect detection of dysplastic cells. Classifier accuracy is typically expressed as the area under the ROC curve (AROC). When the AROC is 1, perfect discrimination is achieved. The LuCED AROC value is 0.991. For single cell detection, the operating point was selected to provide 75% sensitivity and 100% specificity. Cell classification is associated with the detection cases, as shown in the table below. For example, if one aberrant cell is encountered during LuCED analysis, the probability of case detection will be 0.75 or 75%. If two aberrant cells are encountered via LuCED, then the probability of case detection will be (1-(1-0.75) 2 ) = 0.9375 or a sensitivity close to 94% of cases, etc.

[0076] 1 cell - 75% sensitivity,

[0077] 2 cells - 94% case sensitivity, and

[0078] 3 cells - sensitivity in 98% of cases.

[0079] Now refer to Figure 6 , shows an example of a classification cascade for training a classifier suitable for identifying specific mutations associated with different cancer types. Training is performed to produce a series of binary classifiers to separate desired cells, including a first classifier 602, a second classifier 604, a third classifier 608, a fourth classifier 609, a fifth classifier 611, and a sixth classifier 615.

[0080] In one example, training first classifier 602 is to separate malignant cells from other normal cells. First classifier 602 groups all data from malignant cell line and assigns them to a category (e.g., a group of malignant cells). This group of malignant cells plus normal cells are used to train first classifier to separate normal cells from malignant cells as negative controls. Since malignant cells are rare in sputum, this step is particularly important. During training, only a small portion of cells in sputum are manually inspected. Since manual inspection is part of the process, it can be assumed that only the abnormal cells that appear from the process are true malignant cells, and then the classifier described below can be used for subtype classification.

[0081] The second classifier 604 separates malignant subtypes. Any organ system has different types of tissue associated with it. For example, lung tissue includes squamous epithelium and adenoma tissue of the bronchus. Small cell lung cancer (SCLC) cells from the neuroendocrine glands are sometimes also evident. Therefore, a classifier is needed to separate the specific cancer subtype in which the desired driver mutation occurs. This is accomplished by first separating small cell lung cancer from adenocarcinoma and squamous cell carcinoma, and then separating adenocarcinoma from squamous cell carcinoma. Further separation of the desired mutation subtypes in adenocarcinoma is performed step by step. The grouping of cell lines selected as the training set for this example is given in Table 1 below. The separation of specific driver mutations is determined based on the morphological factors in the third to sixth classifiers 608, 609, 611 and 615.

[0082]

[0083] Still refer to Figure 6In one example, mutation-driven stepwise separation begins with a first classifier 602, where a group of cells are separated into normal and malignant or dysplastic categories. Any cells identified as malignant are further processed in a second classifier 604, which separates SCLC: NCI-H69 type cells from other malignant cells and passes them to a third classifier 608. The third classifier 608 separates Adeno: SW900 from other adenocarcinoma type cells and passes them to a fourth classifier 609. The fourth classifier 609 separates Adeno: ALK+, NCI-H2228 cell types from the remaining cell types and passes the remaining cell types to a fifth classifier 611. The fifth classifier 611 separates Adeno: wild type, A549 from EGFR+ adeno cell types and passes the EGFR+ adeno subtype to a fifth classifier 615. The sixth classifier 615 separates adeno:T790M, NCI-H1975 from adeno:EGFR-p.E746_A750del.

[0084] Those skilled in the art will recognize that this is merely one example of an application of the present invention, and that other cell types and mutation drivers can be used to construct and train classifiers, including TMB classifiers, according to the methods described herein. The present invention is in no way limited to this example. Classifier decisions are made by establishing decision boundaries for any measurable characteristic of a feature during the classifier training process. Thresholds can be selected or set based on instrument specifications, acceptable error rates, statistical data, or other criteria based on recognized pattern recognition principles.

[0085] Experimental results

[0086] Now refer to Figure 7 , summarizing the results of an experimental study. Table 650 indicates the area under the ROC (aROC) 652 of target cells classified by a classifier trained according to the above training method, as well as the sensitivity 654 and specificity 656. Specificity is related to the misidentification of malignant cells by a classifier designed to isolate specific driver mutations. For example, the specificity of identifying cells from small cell lung cancer (SCLC) tumors is 99.98%. A small number of cells called SCLC are actually from some other cell lines listed in Table 650. Since only 0.02% of malignant cells are misidentified, the positive predictive value can be calculated to identify SCLC as PPV=TP / (TP+FP)=100*0.748 / (0.748+0.002)=99.7.

[0087] right The excellent discrimination between normal and abnormal cells in the process, coupled with published evidence showing that morphological changes in malignant cells are associated with the cells' genomic features, suggests that it may be possible to identify genetic mutations responsible for driving cancer progression using purely morphological methods. 52 .

[0088] In this disclosure, we extend the concept of morphometric genomics and apply it to the domain of detecting cancer drivers into tumor mutational burden. TM The system is used to generate a morphometric basis for the extent of tumor mutations, with the idea being to provide a non-invasive means of characterizing TMB. Algorithm detects cancer with equal sensitivity regardless of tumor histology, stage, and size 46 Therefore, Cell-CT-based TMB measurement would have the potential to characterize TMB non-invasively, independent of histology, stage, and size factors.

[0089] Now refer to Figure 8 , a diagram showing an embodiment of a method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB). The method includes the following actions:

[0090] Selected clone 830 was obtained from the transduced cells;

[0091] Selected clones were analyzed for MLH1 expression to screen for clones 832 with levels comparable to the parental cell line;

[0092] Selected clones 842 were expanded in culture;

[0093] Selected clone 843 was harvested;

[0094] The TMB levels of the harvested clones were determined 850;

[0095] Analyzing the selected clone 852 on a 3D microscope optical tomography system; and

[0096] Selected clones were compared 854 to a panel of control cell lines.

[0097] In one example, the panel of control cell lines included the parental NCI-H23 and clones expressing scrambled shRNA. Furthermore, the analysis process generates multiple morphometric biomarkers for each cell. TMB data can be determined by genomic analysis using whole-exome sequencing or targeted exome sequencing using NGS or targeted gene panel sequencing.

[0098] Now refer to Figure 9, shows a flowchart of an example of a method for developing one or more morphometric classifiers to use tumor mutation burden (TMB) as ground truth. This method can be combined with the above Figure 8 The method described is used in combination. Low TMB and high TMB can be used as basic facts, for developing a cell classifier for each cell in an isogenic cell line and determining the ROC area 930 of each cell classifier. For each cell in an isogenic cell line, a score 932 that matches the basic facts is defined. By using an adaptive boosted logistic regression algorithm to define a set of projection axes used by a logic function to produce a score from 0 to 1, a classifier is trained for producing a score that closely matches the basic facts. The adaptive boosted logistic regression algorithm can iterate by using the difference between the basic facts and the current score to weight each observed value in a continuous trial to adaptively converge on a solution that gradually converts a set of more extensive cell characterizations into a solution. Alternatively, analyzing selected clones can include using a random forest algorithm to use a non-parametric hypothesis for feature distribution to produce a classifier. In addition, the assessment 950 of the classifier discrimination can be improved by pruning possible feature tree sets to optimize discrimination. The area under the ROC curve, aROC, is used to judge the efficacy of the classifier, where the area under the receiver operating characteristic curve, or aROC, is calculated by calculating the integral of the ROC curve, which represents the overall performance of the binary classifier output in terms of classification sensitivity and specificity 952. A threshold is established for use with the classifier score to create a binary output 954 that is associated with the underlying TBM with high accuracy. A numerical score can be generated that represents the probability that the cell belongs to the target class. By applying a threshold to the distribution to further make the score binary, target cells can be separated from non-target cells. In a useful example, a threshold can be determined to provide an accuracy of 0.95 or higher to separate cells with low TMB from cells with high TMB.

[0099] Now refer to Figure 10 , a flow chart showing an example of a method for treating a malignant tumor in a human subject using immunotherapy. The method includes the following actions:

[0100] analyzing a 3D image of the cell based on pseudo-projections obtained from a sample obtained from the subject 1030;

[0101] operating a biological sample classifier to identify normal or abnormal cells from the sample 1032;

[0102] The TMB score 1042 from each abnormal cell was determined;

[0103] applying a predetermined threshold to the TMB score 1052;

[0104] When cancer is found, surgery is done to remove the cancerous lesion;

[0105] When the TMB score exceeds a predetermined threshold, the human subject is assisted as a candidate for immunotherapy by administering an immunomodulatory agent to the subject for a predetermined period of time to assist the subject's immune system in eliminating cancer cells 1054. The immunomodulatory agent can advantageously be a drug selected from the group consisting of chimeric immune receptors, prostacyclin analogs, iloprost, chimeric antigen receptors (CARs) for T cells, vorinostat, HDAC inhibitors, cholecalciferol, calcitriol, and combinations thereof.

[0106] Data supporting this concept are currently being developed. The approach will include the following actions:

[0107] 1. Bailis et al. 53 (PLoS One. 2013 Oct 29; 8(10): e78726) knocked out the expression of the MLH1 gene in NCI-H23 lung adenocarcinoma cells to generate an isogenic line for direct comparison of MMR-proficient and MMR-deficient cells. Bailis et al. reported that after several weeks in culture, MMR-deficient cells exhibited microsatellite instability, a common mechanism by which cancer cells acquire high TMB. A similar approach can be used by transducing NCI-H23 cells with MLH1 shRNA lentiviral particles (Santa Cruz Biotechnology, sc-35943-V) or scrambled shRNA lentiviral particles (Santa Cruz Biotechnology, sc-108080) and selecting for integration using puromycin. MLH1 shRNA-transduced clones derived from single puromycin-resistant cells can be analyzed by Western blot analysis to screen for clones with >90% reduction in MLH1 expression. Clones can also be derived from cells transduced with control scrambled shRNAs and analyzed for MLH1 expression to screen for clones with levels comparable to the parental NCI-H23 cell line. Selected clones can be cultured and expanded by weekly harvested aliquots and fixed in an ethanol-based fixative for further analysis. Fixed cells can be preliminarily analyzed using FoundationOne analysis (Foundation Medicine, Inc.) to determine TMB levels, which generates a comprehensive genomic profile of more than 300 cancer-related genes. Once cell lines with different TMB levels are identified, they can be analyzed on the VisionGate Cell-CT platform and compared with control cell lines (parental NCI-H23 and clones expressing scrambled shRNA) to verify the accuracy of the identified cells.

[0108] 2. In Cell-CT TM The morphometric features were processed on the 3D microscope optical tomography system platform to generate 704 morphometric biomarkers per cell for each isogenic cell line - this can be accomplished using the 3D microscope optical tomography system platform, image processing and cell classification using the training techniques and functions described in this article.

[0109] 3. Using low and high TMB as ground truth, a cell classifier can be developed for each isogenic cell line and the ROC area determined for each classifier. Typically, this process involves defining a score that matches the ground truth for the cell in question. In this application, the ground truth is high and low TMB, as defined in point 2 above. The classification process aims to produce a score that closely matches the ground truth. Several methods can be used to achieve this, including:

[0110] a. Adaptive Boosted Logistic Regression 50 The method uses principal component projection to define the projection axis, which is then used by the logit function to produce a score between 0 and 1. The algorithm iterates through successive trials that weight each observation by the difference between the ground truth and the current score. This adaptive process converges on a solution that gradually transforms a wider set of cellular representations into a solution.

[0111] b. Random Forest 51 In this method, a nonparametric assumption about the feature distribution is used to generate a classifier. A limitation of adaptive boosting is the assumptions made about the feature distribution behind the principal component processing. This is potentially problematic because features may not strictly conform to the assumed distribution, leading to inaccurate projections. In this method, random vectors of random length are defined. Discrimination is evaluated, and the set of potential feature trees is pruned to optimize discrimination.

[0112] 4. The area under the ROC curve (aROC) can be used to judge the efficacy of a classifier. The area under the receiver operating characteristic curve, or aROC, is calculated by calculating the integral of the ROC curve and represents the overall performance of the binary classifier output in terms of classification sensitivity and specificity. The term "sensitivity" refers to the ability of a classifier to correctly classify objects that have the attribute (or set of attributes) that the classifier is trained to detect as "target" or positive objects. Similarly, specificity represents the ability of a classifier to correctly classify objects that do not have the target attribute as "non-target" or "negative." Both sensitivity and specificity can range from 0 to 1, and it is desirable for a classifier to perform with both parameters as close to 1 as possible. Although a classifier generates a continuous number (a score, i.e., the probability that the object belongs to the target class) as output for each object (in our case, a cell), the output can be further binarized by applying a threshold to the score that separates the positive and negative classes. Once a classifier is developed, an ROC curve can be generated by calculating sensitivity and specificity values as a function of the threshold used to separate the two object classes. aROC values can range from 0 to 1 and represent the percentage of true positive and true negative cells correctly classified by the classifier. In one example, true positive cells are cells with high TMB, while true negative cells are cells with low TMB. Therefore, an aROC > 0.95 means that the classifier correctly classified > 95% of all cells as having low or high TMB.

[0113] 5. Establish a threshold to be used with the classifier score to create a binary output that is highly correlated with the underlying TBM.

[0114] As described in the previous section, a classifier typically produces a numerical score representing the probability that a cell belongs to the target class. To separate target cells from non-target cells, the scores are further binarized by applying a threshold to the score distribution. Typically, scores above the threshold are considered "positive" or target cells, while scores below the threshold are considered "negative" or non-target cells. Because the value of the classifier threshold can vary across the entire range of the score distribution, the ultimate metric for selecting an appropriate value is the classifier's highest accuracy in correctly distinguishing (classifying) cells. In our case, the threshold will be determined to provide an accuracy of 0.95 or higher to separate cells with low TMB from those with high TMB. TMB data is determined by genomic analysis using whole-exome sequencing or targeted exome sequencing using NGS or targeted gene panel sequencing. High and low TMB values are determined based on the TMB distribution. Typically, a TMB above 80% is considered high, although other metrics can also be applied depending on the distribution characteristics.

[0115] Classifier Training - Input and Methods

[0116] The creation and optimization of cell detection classifiers is often referred to as "classifier training" because the process is designed to accurately diagnose cells based on a reference or ground truth. Using the classification methods described herein, cells can be classified into types including, but not limited to, normal, cancerous, and dysplastic. Accuracy has two main aspects: the first is specificity (the classifier calls normal cells normal), and the second is sensitivity (the classifier calls abnormal cells abnormal). Algorithmic training methods include adaptive boosted logistic regression and random forests. Those skilled in the art will be familiar with how to apply other classical training techniques to classifiers, such as template methods, adaptive processing, and the like.

[0117] The method used to train the classifier ensures very good results when the data used as input is used. First, classifier accuracy is ensured when the input to the classifier training process accurately describes clinically relevant aspects of the cells and is robust to environmental factors that may affect the results of the optical tomography system:

[0118] 1.Cell-CT TM The three-dimensional cell images generated by the optical tomography system have high resolution, allowing precise measurement of key features that support correct classification.

[0119] 2. Some features useful in classification only appear in 3D images. Therefore, compared with 2D imaging, 3D feature sets can not only better describe cells but also make classification based on 3D imaging more accurate.

[0120] 3. 3D image segmentation algorithms have been developed to separate whole cells from background and nuclei from cells. The accuracy of these segmentation algorithms has been validated by comparing the segmented traces with human-derived cell or nuclear envelope traces.

[0121] 4. Feature measurements describe various aspects of the cell, nucleus, cytoplasm, and nucleolus. In one example of the tested system, 594 features were calculated for each 3D cell image. These features represent the object's shape, volume, chromatin distribution, and other more subtle morphological elements. The calculation of these features has been shown to be independent of the cell's orientation.

[0122] 5. The diagnostic truth (gold standard in pathology) used for classifier training is typically based on graded cytodiagnoses provided by two cytotechnologists and one cytopathologist.

[0123] Classifier Training - Statistical Considerations

[0124] Second, in a test conducted by the inventors of this paper, the accuracy of the classifier training process was ensured through a rigorous process covering three aspects:

[0125] 1. The database used to train the classifier is formulated to contain enough material to ensure that the binomial 95% confidence interval keeps the variance of the performance estimate within an acceptable range.

[0126] 2. Overtraining is a potential pitfall in the training process where too much information can be incorporated into the classifier, so that it becomes over-specialized for the data used in training. This situation produces an overly optimistic estimate of the classifier's performance. The risk of overtraining can be mitigated through cross-validation, which involves taking a portion of the training data and using it as test data. When the performance estimate based on the training data exceeds the estimate based on the test data, the limit on the amount of information that can be used in the classifier has been reached.

[0127] 3. Finally, as a further guarantee against overtraining, the classifier was tested on data from a second set of cells that were not part of the training process.

[0128] Overview of Abnormal Cell Classifier Training

[0129] The following considerations were used to define the parameters governing the training of the abnormal cell classifier:

[0130] 1. Because abnormal cell samples are rare and non-diagnostic elements are abundant, the classifier must operate with high sensitivity and very high specificity. As shown in Table 2, when the sensitivity of the single cell classifier is 75% and the sample contains more than one abnormal cell, high case detection sensitivity can be maintained.

[0131] 2. To ensure that the workload remains within a reasonable range, the specificity target is set at 99%.

[0132] 3. Lower Binomial 95% Confidence Interval (21) The interval should remain above 70% sensitivity and 98.5% specificity.

[0133] Ultimately, a high detection rate is desirable for each positive case. The sensitivity of single-cell detection translates into the detection of abnormal cases, as shown in Table 2.

[0134] Table 2

[0135]

[0136] The implications of Table 2 are important for lung cancer detection tests. The results shown in this table indicate that if abnormal cells are present in the group analyzed by the lung cancer detection test, they can be detected with confidence, allowing the case to be identified with high sensitivity. The remaining issue is the presence of abnormal cells in the lung cancer detection test analysis, which is the remaining factor that determines the cancer detection rate.

[0137] Classifier Development and Characteristics

[0138] Typically, features are calculated to provide numerical representations of various aspects of a 3D tomogram. The calculated features are used together with expert diagnosis of the object to develop a classifier that can distinguish between object types. For example, a data set with M 3D tomograms and N 3D tomograms calculated for a first type, type 1 object can be calculated for a second type, type 2 object (such as normal and abnormal cells). Here, "M" and "N" represent the number of type 1 values and type 2 values, respectively. The data set is preferably generated by an optical tomography system. The optical tomography system provides a 3D tomography comprising a 3D image of an object (e.g., a cell). Cells typically include other features, such as a nucleus with organelles such as a nucleolus. Object types can include different types of cells, organelles, cells exhibiting a selected disease state, probes, normal cells, or other features of interest, such as features related to TMB. A set of x 3D image features is calculated based on the 3D tomograms of all M+N objects. Next, a refined feature set of y 3D image features that best distinguish object types is found, where "x" and "y" represent the number of 3D image features at each stage. The refined 3D image feature set is used to construct a classifier whose output is correlated with object type. In one exemplary embodiment, at stage 102, a set of 3D tomograms is assembled, where the assembled set represents substantially all of the important markers that an expert would use to distinguish between 3D biological object types. After assembling a representative set of 3D tomograms, a 3D image feature set can be calculated for each object representing the important markers.

[0139] feature

[0140] Tomograms of biological objects such as cells exhibit a variety of observable and measurable features, some of which can be used as features for classification. Table 3 below provides a summary of features, i.e., important markers used to facilitate classification goals.

[0141] Table 3

[0142] feature

[0143]

[0144]

[0145] By way of further explanation, in one useful example, it has been discovered that the presence of voids in a 3D biological object is a useful classification feature based on a measurement criterion, the measurement criterion including comparison to a calculated or selected threshold. Another feature related to voids can include the number of voids in the object. Another feature related to voids can include the volume of the voids or the number of voids. Another feature can include the surface area of the voids or the number of voids. The shape and position of the internuclear voids can also serve as useful features. Additionally, combinations of feature representations can be used to construct classifiers as described above.

[0146] Similarly, it has been found that invaginations in 3D biological objects are useful classification features based on measurement criteria, including comparison with calculated or selected thresholds. Another feature related to invaginations can include the number of invaginations in the object. Another feature related to invaginations includes the volume of invaginations or the number of invaginations. Another feature includes the size of invaginations or the number of invaginations. The location of nuclear invaginations also comprises a useful feature. In addition, combinations of feature representations can also be used to construct classifiers as described above.

[0147] It has been found that invaginations in 3D biological objects are useful classification features based on measurement criteria, including comparison with calculated or selected thresholds. The volume, surface area, shape of the invaginated void, the location of voids connected to the invagination, and combinations of invagination features can also be advantageously used to construct classifiers as described above.

[0148] It has been found that nucleoli present in 3D biological objects are useful classification features based on measurement criteria (including comparison with calculated or selected thresholds). As mentioned above, the volume, surface area, shape, position of objects that may be nucleoli or chromatin condensations, and combinations of the above features can also be advantageously used to construct classifiers. It has now been found that nuclear texture features occurring in 3D biological objects are useful classification features. Using structural elements of various sizes, fuzzy residual techniques can be used to separate features of various sizes within the nucleus. Fuzzy residual techniques typically require blurring the image using a filter and measuring the resulting fuzzy residual by applying a marking operation. The total 3D volume, the number of discrete components, the volume histogram, the mean volume and variance, and the shape histogram are then calculated.

[0149] Distance metrics describing the spatial relationships between nucleoli, invaginations, voids, and the nuclear envelope have been found to be useful classification features. For example, if the mean and variance of three nucleoli are found, the minimum and maximum internucleolar distances can be found. Similarly, the distance between the mean coordinates of a nucleolar cluster and the center of mass of the entire object can be found. Similar calculations can be formed by replacing any of the above entities with the center of mass of the nucleolus and the nucleus.

[0150] It has also been found that Fast Fourier Transform (FFT) features are useful classification features. FFT features are formed by the Fast Fourier Transform of the 3D tomogram. FFT features represent the prominent features and average features of FFT classification.

[0151] The present invention has been described herein in considerable detail in order to comply with patent statutes and to provide those skilled in the art with the information necessary to apply the novel principles of the invention and to construct and use such exemplary and specialized components. However, it should be understood that the invention may be implemented by different equipment and devices, and that various modifications may be made to the details of the equipment and the operating procedures without departing from the true spirit and scope of the invention.

[0152] The disclosures of the following publications are incorporated herein by reference.

[0153] 1. Reddy K., Feinberg A. Higher order chromatin organization in cancer; Seminars in Cancer Biol. 2013:23:109-115.

[0154] 2. Mattout A., Cabianca D., Gasser S., Chromatin states and nuclear organization in development-a view from the nuclear lamina. Genome Biol., 2015; 16:174.

[0155] 3. Zink, D., Fischer, A., Nickerson J. Nuclear structure in cancer cells. Nat. rev.; Cancer. 2004: 4: 677-687.

[0156] 4.Polak, P., Karlic, R., Koren, A., Thurman, R., Sandstrom, R., Lawrence, M., Reynolds, A. et al., Cell-of-origin chromatin organization shapes the mutationallandscape of cancer. Nature, 2015; 518:360-364.

[0157] 5.Schuster-Boeckler B.,Lehner,B.Chromatin organization is a majorinfluence on regional mutation rates in human cancer cells.Nature,2012;488:504-507。

[0158] 6.Pardoll DM.The blockade of immune checkpoints in cancerimmunotherapy.;Nat Rev Cancer.2012 Mar 22;12(4):252-64。

[0159] 7.Robert C,Long GV,Brady B,Dutriaux C,Maio M,Mortier L,etal.Nivolumab in previously untreated melanoma without BRAF mutation.N Engl JMed 2015;372:320-30。

[0160] 8.Weber JS,D'Angelo SP,Minor D,Hodi FS,Gutzmer R,Neyns B,et al.;Nivolumab versus chemotherapy in patients with advanced melanoma whoprogressed after anti-CTLA-4 treatment(CheckMate 037):a randomised,controlled,open-label,phase 3 trial.Lancet Oncol 2015;16:375-84。

[0161] 9.Borghaei H,Paz-Ares L,Horn L,Spigel DR,Steins M,Ready NE,etal.Nivolumab versus docetaxel in advanced nonsquamous non-small-cell lungcancer.N Engl J Med 2015;373:1627-39。

[0162] 10.Gangadhar,Tara C.,Vonderheide,Robert H.Mitigating the toxiceffects of anticancer immunotherapy,Nat.Rev.Clin.Oncol.2014;11(2):91-99。

[0163] 11.Byun,D.,Wolchok,J.,Rosenberg,L.M.,Girotra,M.,Cancer immunotherapy-immune checkpoint blockade and associated endocrinopathies.2017;13:195-207。

[0164] 12.Abdel-Wahab,N.,Alshawa,A.,Suarez-Almazor,M.Adverse effects incancer immunotherapy.Adv.Exp.Med.Biol.,2017;995:155-174。

[0165] 13.Hodi FS,O'Day SJ,McDermott DF,Weber RW,Sosman JA,Haanen JB,etal.Improved survival with ipilimumab in patients with metastatic melanoma.NEngl J Med 2010;363:711-23。

[0166] 14.Motzer RJ,Escudier B,McDermott DF,George S,Hammers HJ,Srinivas S,et al.Nivolumab versus everolimus in advanced renal-cell carcinoma.N Engl JMed 2015;373:1803-13。

[0167] 15.Voong KR,Feliciano J,Becker D,Levy B.Beyond PD-L1 testing-emergingbiomarkers for immunotherapy in non-small cell lung cancer.Ann TranslMed.2017 Sep;5(18):376。

[0168] 16.Danilova L,Wang H,Sunshine J,Kaunitz GJ,Cottrell TR,Xu H,EsandrioJ,Anders RA,Cope L,Pardoll DM,Drake CG1,Taube JM.Association of PD-1 / PD-Laxis expression with cytolytic activity,mutational load,and prognosis inmelanoma and other solid tumors.Proc Natl Acad Sci U S A.2016 Nov 29;1 13(48):E7769-E7777。

[0169] 17.Kefford R.,Ribas A.,Hamid O.,et al.Clinical efficacy andcorrelation with tumor PD-L1 expression in patients with melanoma treatedwith the anti-PD-1 monoclonal antibody MK-3475.Journal of ClinicalOncology.2014;32([supplement:abstr 3005])。

[0170] 18.Carbognin L.,Pilotto S.,Milella M.,et al.Differential activity ofnivolumab,pembrolizumab and MPDL3280A according to the tumor expression ofprogrammed death-ligand-1(PD-L1):sensitivity analysis of trials in melanoma,lung and genitourinary cancers.PLoS ONE.2015;10(6)。

[0171] 19.Muro K.,Bang Y.,Shankaran V.,et al.Relationship between PD-L1expression and clinical outcomes in patients(Pts)with advanced gastric cancertreated with the anti-PD-1 monoclonal antibody pembrolizumab(Pembro;MK-3475)in KEYNOTE-012.Journal of Clinical Oncology.2015;33([supplement;abstr 3]):3-3。

[0172] 20.Madore J.,Vilain R.E.,Menzies A.M.,et al.PD-L1 expression inmelanoma shows marked heterogeneity within and between patients:Implicationsfor anti-PD-1 / PD-L1 clinical trials.Pigment Cell and Melanoma Research.2015;28(3):245-253。

[0173] 21.Mitchell P.,Murone C.,Asadi K.,et al.PD-L1 expression in NSCLC:analysis of a large early stage cohort:and concordance of expression inprimary,node and metastasis.Journal of Thoracic Oncology.2015,10([supplement:S199 abstr]].

[0174] 22.Le DT,Durham JN,Smith KN,Wang H,Bartlett BR,Aulakh LK,Lu S,Kemberling H,Wilt C,Luber BS,Wong F,Azad NS,Rucki AA,Laheru D,Donehower R,Zaheer A,Fisher GA,Crocenzi TS,Lee JJ,Greten TF,Duffy AG,Ciombor KK,EyringAD,Lam BH,Joe A,Kang SP,Holdhoff M,Danilova L,Cope L,Meyer C,Zhou S,GoldbergRM,Armstrong DK,Bever KM,Fader AN,Taube J,Housseau F,Spetzler D,Xiao N,Pardoll DM,Papadopoulos N,Kinzler KW,Eshleman JR,Vogelstein B,Anders RA,DiazLA Jr.Mismatch repair deficiency predicts response of solid tumors to PD-1blockade.Science.2017 Jul 28:357(6349):409-413.

[0175] 23.Viale G,Trapani D,Curigliano G.Mismatch Repair Deficiency as aPredictive Biomarker for Immunotherapy Efficacy.Biomed Res Int.2017;2017:4719194.

[0176] 24.Boussiotis V.A.Somatic mutations and immunotherapy outcome withCTLA-4 blockade in melanoma.New England Journal of Medicine.2014;371(23):2230-2232。

[0177] 25.Goodman AM,Kato S,Bazhenova L,Patel SP,Frampton GM,Miller V,Stephens PJ,Daniels GA,Kurzrock R.Tumor Mutational Burden as an IndependentPredictor of Response to Immunotherapy in Diverse Cancers.Mol CancerTher.2017 Nov;16(11):2598-2608。

[0178] 26.Rizvi N.A.,Hellmann M.D.,Snyder A.Mutational landscape determinessensitivity to PD-1 blockade in non-small cell lung cancer.Science.2015;348(6230):124-128。

[0179] 27.Carbone DP,Reck M,Paz-Ares L,Creelan B,Horn L,Steins M,Felip E,wanden Heuvel MM,Ciuleanu TE,Badin F,Ready N,Hiltermann TJN,Nair S,Juergens R,Peters S,Minenza E,Wrangle JM,Rodriguez-Abreu D,Borghaei H,Blumenschein GRJr,Villaruz LC,Havel L,Krejci J,Corral Jaime J,Chang H,Geese WJ,Bhagavatheeswaran P,Chen AC,Socinski MA;CheckMate 026 Investigators.First-Line Nivolumab in Stage IV or Recurrent Non-Small-Cell Lung Cancer.N Engl JMed.2017 Jun 22:376(25):2415-2426.

[0180] 28.Davis AA,Chae YK*and Giles FJ.Is Circulating Tumor DNA(Ctdna)UseReady For Prime Time?

[0181] 29.Chae YK,Davis AA,Jain S,Santa-Maria C,Flaum L,Beaubier N,PlataniasLC,Gradishar W,Giles FJ,Cristofanilli M.Concordance of Genomic Alterations byNext-Generation Sequencing in Tumor Tissue versus Circulating Tumor DNA inBreast Cancer.Mol Cancer Ther.2017 Jul:16(7):1412-1

[0182] 30.Kim H,Jen J,Vogelstein B,Hamilton SR.Clinical and pathologicalcharacteristics of sporadic colorectal carcinomas with DNA replication errorsin microsatellite sequences.Am J Pathol.1994 Jul;145(1):148-56。

[0183] 31.Ward R,Meagher A,Tomlinson I,O'Connor T,Norrie M,Wu R,HawkinsN.Microsatellite instability and the clinicopathological features of sporadiccolorectal cancer.Gut 2001 Jun;48(6):821-9。

[0184] 32.Greenson JK,Bonner JD,Ben-Yzhak O,Cohen HI,Miselevich I,ResnickMB,Trougouboff P,Tomsho LD,Kim E,Low M,Almog R,Rennert G,Gruber SB.Phenotypeof microsatellite unstable colorectal carcinomas:Well-differentiated andfocally mucinous tumors and the absence of dirty necrosis correlate withmicrosatellite instability.Am J Surg Pathol.2003May;27(5):563-70。

[0185] 33.Alexander J,Watanabe T,Wu TT,Rashid A,Li S,HamiltonSR.Histopathological identification of colon cancer with microsatelliteinstability.Am J Pathol.2001 Feb;158(2):527-35。

[0186] 34.David Gisselsson,Jonas Bjork,Mattias Hoglund,Fredrik Mertens,PaolaDal Cin,Mans Akerman,and Nils Mandahl.Abnormal Nuclear Shape in Solid TumorsReflects Mitotic Instability.Am J Pathol.2001 Jan;158(1):199-206。

[0187] 35.Bai H,Madabushi A,Guan X,Lu AL.Interaction between human mismatchrepair recognition proteins and checkpoint sensor Rad9-Rad1-Hus1.DNA Repair(Amst).2010 May 4;9(5):478-87。

[0188] 36.Debes JD,Sebo TJ,Heemers HV,Kipp BR,Haugen DL,Lohse CM,TindallDJ.p300 modulates nuclear morphology in prostate cancer.Cancer Res.2005 Feb1;65(3):708-12。

[0189] 37.Ionov Y,Matsui S,Cowell JK.A role for p300 / CREB binding proteingenes in promoting cancer progression in colon cancer cell lines withmicrosatellite instability.Proc Natl Acad Sci USA.2004 Feb 3;101(5):1273-8。

[0190] 38.Koshiishi N,Chong JM,Fukasawa T,Ikeno R,Hayashi Y,Funata N,NagaiH,Miyaki M,Matsumoto Y,Fukayama M.p300 gene alterations in intestinal anddiffuse types of gastric carcinoma.Gastric Cancer.2004;7(2):85-90。

[0191] 39.Pihan GA,Purohit A,Wallace J,Malhotra R,Liotta L,DoxseySJ.Centrosome defects can account for cellular and genetic changes thatcharacterize prostate cancer progression.Cancer Res.2001 Mar 1;61(5):2212-9。

[0192] 40.Robinson HM,Black EJ,Brown R,Gillespie DA.DNA mismatch repair andChk1-dependent centrosome amplification in response to DNA alkylationdamage.Cell Cycle.2007 Apr 15;6(8):982-92。

[0193] 41.Bornens M.Cell polarity:intrinsic or externally imposed?New Biol.,3:627-636,1991。

[0194] 42.Rizzolo L.J.,Joshi H.C.Apical orientation of the microtubuleorganizing center and associated g-tubulin during the polarization of theretinal pigment epithelium in vivo.Dev.Biol.,157:147-156,1993。

[0195] 43.Meads T.,Schroer T.A.Polarity and nucleation of microtubules inpolarized epithelial cells.Cell Motil.Cytoskeleton,32:273-288,1995。

[0196] 44.Whitehead C.M.,Salisbury J.L.Regulation and regulatory activitiesof centrosomes.J.Cell.Biochem.,32-33(Suppl.):192-199,1999。

[0197] 45.Meyer M.,Hayenga JW,Neumann T,Katdare R,Presley C,Steinhauer DE,Bell TM,Lancaster CA,Nelson AC.The Cell-CT 3-Dimensional Imaging TechnologyPlatform Enables the Detection of Lung Cancer Using the LuCED SputumTest.Cancer Cytopathology,123(9):512-523,2015。

[0198] 46.Wilbur D.,Meyer MG,Presley C,Aye RW,Zarogoulidis P,Johnson DW,Peled N,Nelson AC.Automated 3-Dimensional Morphologic Analysis of SputumSpecimens for Lung Cancer Detection:Performance Characteristics Support Usein Lung Cancer Screening.Cancer Cytopathology,123(9):548-556,2015。

[0199] 47. Fauver M, Seibel E, Rahn J. Three-dimensional imaging of single isolated cell nuclei using optical projection tomography. Optics Express, 13:4210 - 4223, 2005.

[0200] 48. Deans S. The Radon Transform and Some of Its Applications, New York: Dover Publishers: 2007.

[0201] 49. Rawski [Accessed March 19, 2015]; Maximum intensity projection (MIP) from Radiopaedia.org URL http: / / radiopaedia.org / articles / maximum-intensity- projection-mip obtained.

[0202] 50. Schapire R, Freund Y. Boosting, Foundations and Algorithms, Cambridge: MIT Press, 2012

[0203] 51. Breiman L. Random Forests. Machine Learning, 45:5 - 32, 2001.

[0204] 52. Nelson A., M. Meyer, D. Sussman, R. Katdare, C. Presley, F. Lakers, C. Hamilton, D. Wilbur, R. Mastrangelo, M. Doherty Morphometric Genotyping Identifies Lung Cancer Cells Harboring Target Mutations; Cell-CT Platform Detects Gene Abnormalities, Journal of Thoracic Oncology, 12(11)(Supplement 2): S2278 - S2279, 2017.

[0205] 53.Bailis JM,Gordon ML,Gurgel JL,Komor AC,Barton JK,Kirsch IR.Aninducible,isogenic cancer cell line system for targeting the state ofmismatch repair deficiency.PLoS One.8:e78726,2013。

Claims

1. A method for developing one or more morphometric classifiers to identify tumor mutation burden (TMB), the method comprising: Selected clones were obtained from the transduced cells; analyzing the MLH1 expression of the selected clones to screen for clones with a level comparable to that of the parental cell line; expanding the selected clones in culture; harvesting the selected clones; Determine the TMB levels of the harvested clones; analyzing the selected clones on a 3D microscope optical tomography system, the analysis comprising using low TMB and high TMB as ground truth to develop a cell classifier for each cell in the isogenic cell line and determining a ROC area for each cell classifier; defining, for each cell in the isogenic cell line, a score that matches the ground truth; and The selected clones were compared to a panel of control cell lines.

2. The method according to claim 1, wherein The panel of control cell lines included parental NCI-H23 and clones expressing scrambled shRNA.

3. The method according to claim 1, wherein The act of analyzing selected clones on a 3D microscope optical tomography system generates multiple morphometric biosignatures for each cell.

4. The method of claim 2 further comprising generating a score that closely matches the ground truth.

5. The method according to claim 1, wherein Analyzing the selected clones involves using an adaptive boosted logistic regression algorithm to define a set of projection axes that are used through a logistic function to produce a score from 0 to 1.

6. The method according to claim 5, wherein: The adaptive boosted logistic regression algorithm iterates through successive trials that weight each observation using the difference between the ground truth and the current score to adaptively converge on a solution that gradually transforms a broader set of cellular representations into a solution.

7. The method according to claim 1, wherein Analyzing the selected clones includes using a random forest algorithm to generate a classifier using non-parametric assumptions about feature distributions.

8. The method of claim 7, further comprising evaluating classifier discrimination by pruning the set of potential feature trees to optimize discrimination.

9. The method of claim 1 , further comprising using the area under the ROC curve (aROC) to judge classifier efficacy, wherein the area under the receiver operating characteristic curve (aROC) is calculated by calculating the integral of the ROC curve, which represents the overall performance of the binary classifier output in terms of classification sensitivity and specificity.

10. The method of claim 1, further comprising establishing a threshold for use with the classifier score to create a binary output that correlates with the underlying TMB with high accuracy.

11. The method of claim 1 , further comprising generating a numerical score representing the probability that the cell belongs to a target class.

12. The method of claim 11, further comprising further binarizing the scores by applying a threshold to the score distribution, thereby separating target cells from non-target cells.

13. The method according to claim 10, wherein: The threshold was determined to provide an accuracy of 0.95 or higher for separating cells with low TMB from cells with high TMB.

14. The method of claim 13, further comprising determining TMB data from genomic profiling using whole exome sequencing or targeted exome sequencing using NGS or targeted gene panels.

15. A method for training one or more morphometric classifiers to identify tumor mutation burden (TMB), the method comprising: Selected clones were obtained from the transduced cells; analyzing the MLH1 expression of the selected clones to screen for clones with a level comparable to that of the parental cell line; expanding the selected clones in culture; harvesting the selected clones; Determine TMB level; analyzing the selected clones on a 3D microscope optical tomography system, the analysis comprising using low TMB and high TMB as ground truth to develop a cell classifier for each cell in the isogenic cell line and determining a ROC area for each cell classifier; Multiple morphometric biosignatures were generated for each selected clone; using an adaptive boosted logistic regression algorithm to define a set of projection axes that are used through a logistic function to produce a score from 0 to 1, or using a random forest algorithm to produce a classifier using nonparametric assumptions about the feature distribution; and The selected clones were compared to a panel of control cell lines.

16. The method according to claim 15, wherein The panel of control cell lines included parental NCI-H23 and clones expressing scrambled shRNA.

17. The method of claim 15, further comprising using the area under the ROC curve (aROC) to judge classifier efficacy, wherein the area under the receiver operating characteristic curve or aROC is calculated by calculating the integral of the ROC curve, the integral of the ROC curve representing the overall performance of the binary classifier output in terms of classification sensitivity and specificity.

18. The method of claim 15, further comprising establishing a threshold for use with the classifier score to create a binary output that correlates with the underlying TMB with high accuracy.

19. The method of claim 15, further comprising generating a numerical score representing the probability that the cell belongs to the target class.

20. The method of claim 15, further comprising further binarizing the scores by applying a threshold to the score distribution to separate target cells from non-target cells.

Images