Platelet diagnosis by platelet force pattern analysis

Abstract

A method of assessing platelet health conditions and disorders in a subject is provided. The method involves obtaining a blood sample from the subject, isolating platelets from the blood sample, plating at least a portion of the platelets on a force-sensitive substrate, imaging the platelets on the force-sensitive substrate using fluorescence microscopy to detect platelet adhesive force patterns, and analyzing the platelet adhesive force patterns to produce analytical results. The analytical results are used to identify either normal function or abnormality of the subject's platelets.

Description

PLATELET DIAGNOSIS BY PLATELET FORCE PATTERNANALYSISCROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Application Serial No. 63 / 712,556, filed October 28, 2024, which application is hereby incorporated by reference in its entirety.TECHNICAL FIELD

[0002] The present invention relates to methods of diagnosing platelet disorders.STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

[0003] This invention was made with government support under GM128747 awarded by the National Institute of Health. The U.S. Government has certain rights in the invention.BACKGROUND OF THE INVENTION

[0004] Platelets, also known as thrombocytes, are minute blood cells essential for the blood clotting process. Platelet disorders in patients can have serious consequences, including increased bleeding risks, heightened clotting tendencies (thrombocytosis), susceptibility to infections, and potential organ damage. Hematologists frequently employ platelet functional tests to evaluate the efficacy of a patient's platelets. However, current methodologies, such as platelet aggregometry and flow cytometry, primarily examine platelet-secreted chemicals or measure changes in absorbance caused by platelet aggregation. These methods suffer from a notable rate of falsenegative or false-positive results. Furthermore, they typically demand a significant blood volume (>5 mb), necessitating venous blood draw.SUMMARY OF THE INVENTION

[0005] Certain exemplary aspects of the invention are set forth below. It should be understood that these aspects are presented merely to provide the reader with a brief summary of certain formsthe invention might take and that these aspects are not intended to limit the scope of the invention. Indeed, the invention may encompass a variety of aspects that may not be explicitly set forth below.

[0006] In one aspect of the present invention, a method of assessing platelet health conditions and disorders in a subject is provided. The method involves obtaining a blood sample from the subject, isolating platelets from the blood sample, plating at least a portion of the platelets on a forcesensitive substrate, imaging the platelets on the force-sensitive substrate using fluorescence microscopy to detect platelet adhesive force patterns, and analyzing the platelet adhesive force patterns to produce analytical results. The analytical results are used to identify either normal function or abnormality of the subject’s platelets. In one embodiment, the blood sample is about 10 pL or less.

[0007] In another embodiment, the analysis of the platelet adhesive force patterns comprises the use of an artificial intelligence program. In one embodiment, the artificial intelligence program comprises a machine learning model that is trained for image recognition. In another embodiment, the machine learning program can report platelet conditions of the subject by assessing the platelet force patterns.

[0008] In one embodiment, the artificial intelligence program comprises a convolutional neural network. In another embodiment, the convolutional neural network is a non-transitory computer- readable storage medium comprising program instructions for causing a computer to perform a method of reporting platelet conditions of a subject using a machine learning program that has been trained with platelet force patterns that are predefined as either normal or abnormal.

[0009] In one embodiment, at least about 50 normal platelet force patterns and at least about 50 abnormal platelet force patterns are used to train the machine learning program. In another embodiment, at least 100 normal platelet force patterns and at least 100 abnormal platelet force patterns are used to train the machine learning program. In one embodiment, at least 200 normal platelet force patterns and at least 200 abnormal platelet force patterns are used to train the machine learning program.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The foregoing summary, as well as the following detailed description of preferred embodiments of the application, will be better understood when read in conjunction with the appended drawings.

[0011] FIG. 1 A is a schematic showing a double-stranded DNA-based tension sensor, decorated by a quencher, a dye and a ligand, attracts platelet adhesion.

[0012] FIG. IB is a schematic showing the process of acquiring platelet adhesive force patterns.

[0013] FIG. 1C is the brightfield imaging of platelet adhesive force patterns.

[0014] FIG. ID is the fluorescent image of the platelet adhesive forces.

[0015] FIG. IE is the merged image of the platelet adhesive forces.

[0016] FIG. IF is an enlarged image of individual platelet force patterns.

[0017] FIG. 1G is a graph showing calibrated spatial resolution of platelet adhesive force imaging.

[0018] FIG. 2A is a series of images showing platelet force patterns of healthy platelets versus those of abnormal platelets (treated by anti-platelet drugs).

[0019] FIG. 2B is a photo of an imaging system for platelet adhesive force acquisition.

[0020] FIG. 3A is an excerpt of code for a program that was trained by being exposed to normal platelet force patterns and abnormal platelet force patterns (treated with tirofiban, an anti-platelet drug).

[0021] FIG. 3B is a graph showing 90% accuracy of the training process to identify a normal platelet based on a single platelet force pattern.

[0022] FIG. 3C is a pair of images showing the typical healthy and non-healthy platelet force patterns that are fed to the program for training.

[0023] FIG. 4 is an example of platelet health condition evaluation by the Al.DEFINITIONS

[0024] As used herein, the term “convolutional neural network” means a type of deep learning artificial intelligence specifically designed to analyze data with a grid-like structure, such as images, by automatically and hierarchically extracting features. It uses special layers, including convolutional and pooling layers, to identify patterns in data, from simple features like edges to more complex ones like entire objects. This makes them highly effective for tasks like image recognition, object detection, and medical image analysis.DETAILED DESCRIPTION OF THE INVENTION

[0025] The details of one or more embodiments of the disclosed subject matter are set forth in this document. Modifications to embodiments described in this document, and other embodiments, will be evident to those of ordinary skill in the art after a study of the information provided herein.

[0026] The present disclosure may be understood more readily by reference to the following detailed description of the embodiments taken in connection with the accompanying drawing figures, which form a part of this disclosure. It is to be understood that this application is not limited to the specific devices, methods, conditions or parameters described and / or shown herein, and that the terminology used herein is for the purpose of describing particular embodiments by way of example only and is not intended to be limiting. Also, in some embodiments, as used in the specification and including the appended claims, the singular forms “a,” “an,” and “the” include the plural, and reference to a particular numerical value includes at least that particular value, unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” or “approximately” one particular value and / or to “about” or “approximately” another particular value. When such a range is expressed, another embodiment includes from the one particular value and / or to the other particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another embodiment.

[0027] It should be understood that every maximum numerical limitation given throughout this specification includes every lower numerical limitation, as if such lower numerical limitations were expressly written herein. Every minimum numerical limitation given throughout this specification will include every higher numerical limitation, as if such higher numerical limitations were expressly written herein. Every numerical range given throughout this specification will include every narrower numerical range that falls within such broader numerical range, as if such narrower numerical ranges were all expressly written herein.

[0028] The present invention involves a novel method to test platelet function. As blood clotting is a biomechanical process, normal platelets produce consistent adhesive forces during platelet adhesion, the initial stage of hemostasis. These forces are imaged by plating platelets on a forcesensing substrate and using fluorescence microscopy (FIGs 1A and IB). Normal platelets are shown to display consistent and unified force patterns, while abnormal platelets display altered force patterns or reduced force-caused fluorescence intensities (FIGs 2A and 2B). Using thisdiscovery, the present invention discloses a novel method to diagnose platelet health conditions using an analysis tool such as an artificial intelligence (Al) program trained by machine learning that reads and assesses platelet adhesive force patterns (FIGs 3 A and 3B). An additional advantage of the present invention is that the method only requires a small amount of blood, in the range of 10 pL to 100 pL, which can be provided by fingertip pricking. This is a less traumatic procedure in comparison to a venous blood draw.Force-sensitive substrate

[0029] The force-sensitive substrate is a rigid surface, such as glass, that emits fluorescence in response to contractile forces generated by platelets, allowing visualization of platelet forces through fluorescence imaging. In one embodiment, the force-sensitive substrate is a glass-bottom Petri dish densely grafted with molecular force sensors (ITS, as described below), with a density exceeding 1,000 sensors per square micron. When platelets adhere to this surface, they engage with the force sensors, activating them into a fluorescent state. Imaging these activated sensors produces a spatial map of platelet-generated forces.Platelet force

[0030] To achieve high sensitivity and specificity, the method described in this invention quantifies platelet force as the key indicator for evaluating platelet health. Force plays an important role in virtually all aspects of platelet functions. Following an injury, platelets initially adhere to the exposed sub-endothelium through integrins, which transmit force to mediate stable platelet adhesion and activation. The activated platelets recruit additional platelets to form thrombi and clots. Platelets further contract in the thrombi and clots to stabilize the structures, and eventually stop bleeding. During the whole process, the force transmitted by integrins (integrin a2pi and integrin allb[33 ) play critical roles in platelet adhesion, activation, aggregation and contraction. Therefore, integrin-transmitted force is essential in normal platelet functions. The force-sensing substrate described in this invention show platelet aggregation force by fluorescence, indicating platelet health condition with high sensitivity and specificity.

[0031] The force-sensing substrates of the present invention enable platelet aggregation force examination. Platelet adhesive forces can be detected and imaged on planar glass surfaces using a tension sensor (named integrative tension sensor, or ITS), as shown in FIGs 1 A and IB. ITS emitsfluorescence if a molecular force generated by platelets ruptures the double-stranded DNA structure of the ITS and frees a fluorophore on the ITS from quenching, hence converting force signal to fluorescence signal. The structure and working principle of ITS is shown in FIG. 1 A. The procedure of platelet adhesive imaging is shown in FIG. IB. FIGs 1C-1F show typical platelet force images after platelet plating on ITS surfaces.Integrative Tension Sensor

[0032] An Integrative Tension Sensor (ITS) directly converts molecular tensions to fluorescent signals (force-to-fluorescence conversion). As a molecular linker, the ITS is initially non- fluorescent and can be permanently activated to fluoresce by tension. The tension threshold required for ITS activation is tunable in the range of 10-60 piconewton (pN), therefore enabling the selective visualization of molecular tensions at different force levels. On a surface where integrin ligands are tethered by the ITS, integrin tensions in platelets activate the ITS and can be directly mapped by fluorescence imaging without the need of post modeling and computation which is usually required by force-to-strain approaches.

[0033] ITS is a double stranded DNA (dsDNA) with 18 base pairs. The sequences and modifications of embodiments of the single-stranded DNAs for ITS synthesis are shown below. ThioMC6-D / denotes the thiol conjugation to the 5’ end of DNA. / BiosG / and / Bio / denote the DNA modifications with biotin conjugations at 5’ end and 3’ end, respectively. BHQ2 is the black hole quencher. The DNA sequences were selected and analyzed by a DNA analysis tool available from the company Integrated DNA Technologies (IDTDNA) to minimize the probability of selfdimer and hairpin formation in the ssDNA.

[0034] 5’- / ThioMC6-D / SEQ. ID 1 / BHQ2 / -3’

[0035] SEQ. ID 1 : GGG CGG CGA CCT CAG CAT

[0036] 5’- / BiosG / T / iCy3 / SEQ. ID 2 / -3’

[0037] SEQ. ID 2: ATG CTG AGG TCG CCG CCC

[0038] 5’- / Cy3 / SEQ. ID 2 / Bio / -3’

[0039] 5’- / Alexa647 / SEQ. ID 2 / Bio / -3’

[0040] Cyclic peptide RGD with an amine group (PCI-3696-PI, Peptides International) was conjugated with the DNA strand with thiol modification. The Cyclic RGD has a PEG linker whichenhance the accessibility of the RGD to integrins on cell membrane. Conjugation was conducted according to a previously published protocol. Briefly, the thiol modification on 5’ end of DNA was deprotected by TCEP (Tris(2-carboxyethyl)phosphine hydrochloride) and reacted with the maleimide group of a heterolinker SMCC-sulfo (22622, thermos scientific). The other end of SMCC-sulfo is an NHS ester (N-hydroxysuccinimide esters) which reacts with the amine of RGD. The RGD conjugated DNA was purified by electrophoresis. The 12 pN and 54 pN ITSs used in this paper were assembled by hybridizing the two ssDNAs with a concentration ratio of 1.1 : 1 (The strand with quencher: the DNA strand with fluorophore).

[0041] ITS with 54 pN threshold (with Cy3-BHQ2 pair)

[0042] 5’- / RGD / SEQ. ID 1 / BHQ2 / -3’

[0043] 5’- / BiosG / T / iCy3 / SEQ. ID 2 / -3’

[0044] ITS with 12 pN threshold (with Cy3-BHQ2 pair)

[0045] 5’- / RGD / SEQ. ID 1 / BHQ2 / -3’

[0046] 5’- / Cy3 / SEQ. ID 2 / Bio / -3’

[0047] ITS with 12 pN threshold (with Alexa647-BHQ2 pair)

[0048] 5’- / RGD / SEQ. ID 1 / BHQ2 / -3’

[0049] 5’- / Alexa647 / SEQ. ID 2 / Bio / -3’Al program

[0050] In one embodiment, an Al program is trained with a large amount of predefined individual platelet force patterns produced by healthy (aka “normal”) platelets and unhealthy (aka “abnormal”) platelets under various conditions. Multiple groups of platelets, healthy ones, ones exposed to different anti-platelet drugs, and ones exhibiting known disorders, were tested, and produced force patterns which were subsequently used to train the Al program. Each group of platelets consists of 103-104force images. After being exposed to these databases, this machine learning-trained program developed the ability to automatically evaluate the platelet health conditions by assessing the platelet force patterns (FIG. 4). The Al program is also able to diagnose the platelet diseases, either congenital or acquired, by looking at the platelet force patterns. This program automates the process of platelet diagnosis, eliminates human mistakes and reduces workloads during platelet function tests.

[0051] In one embodiment, the AT program is a self-compiled program script running in Python. Some basic coding modules were imported from Pytorch, an open-sourced Python library for machine learning training. In one embodiment, the Al program is a convolutional neural network, a type of deep learning algorithm designed for processing and categorizing image data. This program instructs the computer to automatically learn the differences of image features between normal platelet force maps and abnormal platelet force maps. In one embodiment, the convolutional neural network is a non -transitory computer-readable storage medium comprising program instructions for causing a computer to perform a method of reporting platelet conditions of a subject using a machine learning program that has been trained with platelet force patterns that are predefined as either normal or abnormal. In one embodiment, “abnormal platelets” used for training are platelets treated with anti-platelet drugs. An example of an anti-platelet drug is tirofiban.

[0052] In one embodiment of the program training, 240 normal platelet force maps and 240 abnormal platelet force maps were used, respectively. For example, users input 240 normal platelet force maps into the computer, labeling them as originating from healthy platelets, and another 240 abnormal force maps labeled as derived from platelets treated with anti-platelet drugs. The program then compares the image features of the two groups and automatically extracts and stores the key features that distinguish healthy platelets from drug-treated (unhealthy) ones.EXAMPLESExample 1

[0053] A force-sensing substrate (typically on a glass) was used to record platelet adhesive force patterns. FIG. 1 A is a schematic showing a double-stranded DNA-based tension sensor, decorated by a quencher, a dye and a ligand, attracts platelet adhesion. The double-stranded DNA is separated by the platelet adhesive force and becomes fluorescent, therefore converting force signal to fluorescent signal. FIG. IB shows the process of acquiring platelet adhesive force patterns. One example of platelet adhesive force patterns is shown in FIGs 1C-1E. FIG. 1C shows the brightfield imaging of platelets. FIG. ID is the fluorescent image of the platelet adhesive forces, and FIG. IE is the merged image. FIG. IF shows one enlarged image of individual platelet force patterns. FIG. 1G is a graph showing the calibrated spatial resolution of platelet adhesive force imaging.Example 2

[0054] Platelets were treated with anti-platelet drugs. They displayed altered adhesive force patterns. FIG. 2A shows platelet force patterns of healthy platelets versus those of abnormal platelets (treated by anti-platelet drugs). The distinct pattern difference between normal platelet force patterns and abnormal ones provides the basis for image-based diagnosis of platelet disorders. FIG. 2B is a photo of the imaging system used for platelet adhesive force acquisition.Example 3

[0055] A machine learning model was trained to identify healthy platelets by platelet force pattern recognition. Referring to FIG. 3 A, a program was trained by being exposed to normal platelet force patterns and abnormal platelet force patterns (treated with tirofiban, an anti-platelet drug). The figure shows an excerpt of the code. The graph of results from the training process in FIG. 3B shows 90% accuracy to identify a normal platelet based on a single platelet force pattern. As one blood sample may produce millions platelet force patterns. By obtaining a consensus from multiple platelet force patterns within a single patient, it is possible to further improve the accuracy of identifying platelet disorders in this patient. FIG. 3C shows the typical healthy and non-healthy platelet force patterns that are fed to the program for training. Tirofiban is a commonly used antiplatelet drug.Example 4

[0056] An example of platelet health condition evaluation by the Al (the machine learning trained program) is shown in FIG. 4. Upper class letters (“TREATED” or “HEALTHY”) are the results reported by program which read the corresponding platelet force patterns. Lower class letters (“treated” or “healthy”) indicate the ground truths which are known to the researchers, but not to the program. Among the 16 randomly selected platelets, 15 platelets are correctly identified (except the circled one).

[0057] All documents cited are incorporated herein by reference; the citation of any document is not to be construed as an admission that it is prior art with respect to the present invention.

[0058] It is to be further understood that where descriptions of various embodiments use the term “comprising,” and / or “including” those skilled in the art would understand that in some specificinstances, an embodiment can be alternatively described using language "consisting essentially of’ or "consisting of.”

[0059] While particular embodiments of the present invention have been illustrated and described, it would be obvious to one skilled in the art that various other changes and modifications can be made without departing from the spirit and scope of the invention. It is therefore intended to cover in the appended claims all such changes and modifications that are within the scope of this invention.SEQUENCES

[0060] SEQ. ID 1 :GGG CGG CGA CCT CAG CAT

[0061] SEQ. ID 2:ATG CTG AGG TCG CCG CCCINCORPORATION BY REFERENCE

[0062] The Sequence Listing associated with this application is filed in electronic format via EFS- Web and is hereby incorporated by reference into the specification in its entirety. The name of the text file containing the Sequence Listing is 2024020.xml. The date of creation of the Sequence Listing is October 28, 2025, and the size of the file in bytes is 3 kb.

Claims

What is claimed is:

1. A method of assessing platelet health conditions and disorders in a subject, the method comprising: a. obtaining a blood sample from the subject; b. isolating platelets from the blood sample; c. plating at least a portion of the platelets on a force-sensitive substrate; d. imaging the platelets on the force-sensitive substrate using fluorescence microscopy to detect platelet adhesive force patterns; and e. analyzing the platelet adhesive force patterns to produce analytical results, wherein the analytical results are used to identify either normal function or abnormality of the subject’s platelets.

2. The method of claim 1 wherein the blood sample is about 10 pL or less.

3. The method of claim 1 wherein the analysis of the platelet adhesive force patterns comprises the use of an artificial intelligence program.

4. The method of claim 3 wherein the artificial intelligence program comprises a machine learning model that is trained for image recognition.

5. The method of claim 4 wherein the machine learning program can report platelet conditions of the subject by assessing the platelet force patterns.

6. The method of claim 4 wherein the artificial intelligence program comprises a convolutional neural network.

7. The method of claim 6 wherein the convolutional neural network is a non-transitory computer-readable storage medium comprising program instructions for causing a computer to perform a method of reporting platelet conditions of a subject using a machine learning program that has been trained with platelet force patterns that are predefined as either normal or abnormal.

8. The method of claim 7 wherein at least about 50 normal platelet force patterns and at least about 50 abnormal platelet force patterns are used to train the machine learning program.

9. The method of claim 7 wherein at least 100 normal platelet force patterns and at least 100 abnormal platelet force patterns are used to train the machine learning program.

10. The method of claim 7 wherein at least 200 normal platelet force patterns and at least 200 abnormal platelet force patterns are used to train the machine learning program.

Images