Methods of monitoring cancer using fragmented profiles
By scoring and machine learning model prediction of free DNA fragmentation maps, the problem of difficult to predict and monitor cancer ctDNA load in the prior art is solved, and the accurate prediction and therapeutic response of cancer progression and survival are achieved, providing A non-invasive biomarker.
Patent Information
- Application Number
- CN202380027237.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-03-17
- Filing Date
- 2023-03-17
- Publication Date
- 2024-11-19
AI Technical Summary
The prior art is difficult to effectively predict and monitor the tumor-derived DNA module (ctDNA) load of cancer, affecting cancer diagnosis, treatment response and patient survival prediction.
By scoring free DNA (cfDNA) fragmentation maps, machine learning models are used to predict cancer progression and survival, combining the ratio of short-to-long fragments, fragment size distribution and model weights, monitoring scores are generated to indicate tumor origin in cfDNA Nucleic acid level.
Improves the accuracy of cancer diagnosis and the ability to predict treatment responses, enhances the prediction of patient survival, and provides a non-invasive, minimally invasive biomarker capable of monitoring cancer progression and treatment response.
Smart Images

Figure CN118984879A_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of U.S. Provisional Patent Application Serial No. 63 / 320,906, filed on March 17, 2022, the entire contents of which are incorporated herein by reference in their entirety. Technical Field
[0003] The present invention relates generally to genetic analysis, and more particularly to a method and system for analyzing cell-free DNA (cfDNA) fragments to predict the fraction of tumor-derived DNA modules (ctDNA burden) and detect cancer in a subject. Background Art
[0004] Most of the morbidity and mortality of human cancers worldwide is due to late diagnosis of these diseases, when treatment is less effective. In addition, after cancer is diagnosed, it is challenging to predict cancer progression and patient response to treatment, which further reduces the success rate of cancer treatment. Unfortunately, clinically proven biomarkers that can be used to widely diagnose patients with cancer and predict effective treatment for said patients are not widely available.
[0005] The fraction (ctDNA load) of tumor-derived DNA molecules in plasma is a tool that can be used to describe the total tumor load of patients with cancer.Previous work shows that the ctDNA load of individual patients is affected by many factors, including the origin tissue and stage of the tumor and vascularization and perfusion.Therefore, the ctDNA load of patients with advanced cancer is higher than the ctDNA load of patients with early cancer.Similarly, the ctDNA load of patients with cancer (such as colorectal cancer) with high cell turnover rate in tissue and directly entering the bloodstream is usually higher than the ctDNA load of patients with less blood vessels and slower growing tumors.ctDNA load may change over time as tumors are exposed to treatment and die (reduced) and subsequently acquire resistance mechanisms to treatment and grow (improve).However, previous studies lack the ability to effectively predict ctDNA load and use the predicted ctDNA load as a tool for diagnosing cancer, predicting disease progression and treatment response and determining the total survival of patients diagnosed with cancer. Summary of the invention
[0006] The present disclosure provides methods and systems for monitoring cancer progression and predicting the overall survival of a subject by scoring a cfDNA fragmentation profile obtained by analyzing cfDNA fragments in a sample obtained from a subject. The scoring method generates features that can be used to train a machine learning model to predict biomarkers that can be used to monitor cancer progression, evaluate a patient's response to treatment, and predict the overall survival of a subject.
[0007] Thus, in one embodiment, the present invention provides a method for monitoring cancer. The method comprises:
[0008] determining a cell-free DNA (cfDNA) fragmentation profile of a sample from a subject;
[0009] calculating a score based on the cfDNA fragmentation profile, the score indicating the likelihood of the presence of cancer in the subject;
[0010] determining the ratio of short fragments to long fragments and the fragment size distribution according to the fragmentation profile;
[0011] calculating a divergence score based on the ratio of the short fragments to the long fragments;
[0012] determining a set of model weights based on the fragment size distribution;
[0013] training a machine learning model using a collection of features extracted from a plurality of fragmentation profiles from a plurality of subjects; and
[0014] Based on the monitoring score, the divergence score and the model weight, a monitoring score for the sample is determined by the machine learning model, wherein the monitoring score indicates the level of tumor-derived nucleic acid in the cfDNA of the sample.
[0015] In some aspects, the cfDNA fragmentation profile is determined by: obtaining cfDNA fragments from the subject and separating the cfDNA fragments; sequencing the cfDNA fragments to obtain sequenced fragments; mapping the sequenced fragments to the genome to obtain a window of mapped sequences; and analyzing the window of mapped sequences to determine the cfDNA fragment lengths and generate the cfDNA fragmentation profile.
[0016] In another embodiment, the present invention provides a method of determining at least one of overall survival, progression-free survival, or time until progression in a subject with cancer, comprising:
[0017] determining a cell-free DNA (cfDNA) fragmentation profile of a sample from the subject;
[0018] calculating a score based on the cfDNA fragmentation profile, the score indicating the likelihood of the presence of cancer in the subject;
[0019] determining the ratio of short fragments to long fragments and the fragment size distribution according to the fragmentation profile;
[0020] calculating a divergence score based on the ratio of the short fragments to the long fragments;
[0021] determining a set of model weights based on the fragment size distribution;
[0022] Training a machine learning model using a collection of features extracted from fragmented atlases of multiple subjects;
[0023] Based on the score, the divergence score, and the model weight, determining a monitoring score for the sample by the machine learning model, thereby indicating the possibility of cancer progression in the subject, the monitoring score indicating the level of tumor-derived nucleic acid in the cfDNA of the sample; and
[0024] Determining at least one of the probability of overall survival, progression-free survival, or time until progression of the subject based on the monitoring score, thereby determining the overall survival of the subject. In yet another aspect, the present invention provides a method of treating a subject having cancer.
[0025] In yet another embodiment, the present invention provides a system for monitoring cancer in a subject. The system comprises:
[0026] Memory; and
[0027] One or more processors coupled to the memory, the one or more processors configured to execute operations that cause the computer system to:
[0028] determining a cell-free DNA (cfDNA) fragmentation profile of a sample from the subject;
[0029] calculating a score based on the cfDNA fragmentation profile, the score indicating the likelihood of the presence of cancer in the subject;
[0030] determining the ratio of short fragments to long fragments and the fragment size distribution according to the fragmentation profile;
[0031] calculating a divergence score based on the ratio of the short fragments to the long fragments;
[0032] determining a set of model weights based on the fragment size distribution;
[0033] training a machine learning model using a collection of features extracted from the fragmented atlases of multiple subjects; and
[0034] Based on the score, the divergence score and the model weight, a monitoring score for the sample is determined by the machine learning model, thereby indicating the possibility of cancer progression in the subject, and the monitoring score indicates the level of tumor-derived nucleic acid in the cfDNA of the sample.
[0035] In another embodiment, the present invention provides a non-transitory computer-readable storage medium encoded with a computer program. The computer program includes instructions that, when executed by one or more processors, cause the one or more processors to perform operations for performing the method of the present invention.
[0036] In yet another embodiment, the present invention provides a computing system, wherein the system comprises a memory and one or more processors coupled to the memory, wherein the one or more processors are configured to execute operations implementing the method of the present invention.
[0037] In yet another embodiment, the present invention provides a system for genetic analysis and assessment of cancer, the system comprising: (a) a sequencer configured to generate a whole genome sequencing (WGS) dataset for a sample; and (b) a non-transitory computer-readable storage medium and / or a computer system of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 is a block diagram illustrating a process for training a machine learning model to generate a DELFI Monitoring Score (DMS). The DMS can be used to monitor cancer in patients diagnosed with cancer, determine overall survival, determine progression-free survival, and determine time until progression. The DMS can also be used to diagnose a patient's cancer and determine a cancer treatment to administer to the patient.
[0039] Figure 2 is a graphical plot showing a comparison between the observed MAF trajectory and the trajectory determined based on the DMS value.
[0040] Figure 3 is a graphical plot showing a comparison between the MAF trajectory observed without reference to the time point or patient from which a given sample originated and the trajectory determined based on the DMS values.
[0041] Figure 4is a graphical diagram showing a comparison between observed MAF trajectories and trajectories determined based on DMS values. The patient cohort with observed MAF trajectories in this figure has a different cancer type than the cancer type of the patient cohort used to train the model to generate DMS values.
[0042] Figure 5 is a graphical diagram showing progression-free survival data obtained by predicted DMS for two different cohorts with different types of cancer.
[0043] Figure 6 is a graphical diagram showing the overall survival of a group of cancer patients at a pre-treatment time point using the predicted DMS. The graphical diagram also shows the overall survival of a group of cancer patients undergoing incomplete resection and complete resection.
[0044] Figure 7 is a graphical diagram showing progression-free survival for a cohort of cancer patients whose clonal variant MAF reached an optimal 0% when assessed after first treatment. The graphical diagram also shows a comparison between the predicted DMS and the MAF measured by ddPCR.
[0045] Figure 8 is an example computer 800 that can be used to implement Figure 1 The training algorithm shown here generates DMS values.
[0046] Figures 9A-9B Study design and patient flow diagram. Fig.9A It is CAIRO5 research design flow chart.After confirming eligibility (including the unresectable state of liver metastasis defined by the central team), KRAS (exon 2, 3 and 4), NRAS (exon 2 and 3) and BRAF mutation status in tissue samples are evaluated. Patients with RAS / BRAF mutant tumors are randomly assigned between double chemotherapy plus bevacizumab (bevacizumab) (the 1st group) or triple chemotherapy plus bevacizumab (the 2nd group). Patients with RAS / BRAF wild-type tumors are randomly assigned between double chemotherapy plus bevacizumab (the 3rd group) or panitumumab (panitumumab) (the 4th group). In this translational study, blood draws from patients in the 1st group (mutant group) and the 3rd group (wild-type group) are processed and analyzed. Fig. 9B is a patient flow diagram. It depicts the number of patients and samples included in the study and the reasons for exclusion.
[0047] Figures 10A-10D DELFI Tumor Fraction (DELFI-TF) is a mutation-agnostic method for metastatic disease monitoring. Fig. 10ATumors from patients with untreated inoperable liver-only mCRC enrolled in the CAIRO5 Phase III trial were tested for hotspot mutations in KRAS, NRAS, and BRAF. Blood samples were collected at baseline, during treatment, and at disease progression or at last follow-up. Patients with driver mutations were monitored using ddPCR and the DELFI-TF assay. Patients with wild-type KRAS, NRAS, and BRAF were monitored with DELFI-TF alone. Fig. 10B It is shown that plasma aliquots from patients with tissue-confirmed RAS / BRAF mutant mCRC are used for cfDNA isolation. From each time point, replicate cfDNA samples are used for ddPCR and low coverage WGS. The WGS fragment sequencing statistics for each sample are calculated at a given time point. A Bayesian probabilistic model is trained for the MAF called by ddPCR readings of tumor-specific RAS / BRAF variants in all longitudinal cfDNA samples to generate DELFI-TF values. Fig. 10C Figure 2 is a heat map of genomic features depicting the deviations in cfDNA fragmentation ratios and chromosome arm-level z-scores across baseline and on-treatment time points as well as DELFI-TF values and clinical and demographic characteristics for 128 patients. Fig. 10D We demonstrated that genome-wide fragmentation profiles of cfDNA in 504 non-overlapping 5-Mb genomic regions at baseline and at time points close to the second imaging assessment by RECIST 1.1 showed significant heterogeneity at baseline and in patients who presented with disease progression compared with those who experienced stable disease or radiation response after initial first-line systemic therapy.
[0048] Figures 11A-11F showed how DELFI-TF predicts tumor fraction in the blood of patients with advanced disease receiving systemic therapy. Fig.11A It is shown that patients with mCRC exhibited a wider range of DELFI-TF values at baseline than those of non-cancer controls. The upper limit of the 95% CI in non-cancer controls (grey dashed line) represents the DELFI-TF limit of the null (0.006). Fig. 11B It is shown that DELFI-TF is strongly correlated with MAF measured by ddPCR at all time points studied (n=692, Pearson correlation, r=0.85, p=3.9e-89). Plasma time points where MAF was undetectable (n=60, red) exhibited a wide range of DELFI-TF values. Fig. 11CFigure 3 shows that cfDNA fragmentation profiles of patients with wild-type mCRC (bottom) exhibit short-to-long ratio distortions, even in the context of tumor copy-neutral regions across 100 kb bins in matched tissue samples (top). Fig.11D Figure 3: Plasma tumor fraction assessed by DELFI-TF (blue) and RAS / BRAF MAF (orange) values correlated with cfDNA copy number changes in genomic regions harboring frequently deleted (MBD1) or amplified (PLGC1) genes in colorectal cancer. Fig.11E and Fig.11F The relative coverage at TSS positions of a panel of 890 genes highly expressed in colorectal cancer shows significantly deeper valleys in baseline samples (brown) than in on-treatment samples (purple) (Wilcoxon test, p<0.001), indicating that, on average, tumor scores are lower at time points after the start of treatment.
[0049] Figures 12A-12C Shown is the cfDNA tumor fraction assessed by whole-genome sequencing approach. Fig. 12A It is shown that patients with metastatic colorectal cancer exhibit a wider range of ichorCNA values at baseline than the range of ichorCNA values of non-cancer controls. The upper limit of the 95% confidence interval for non-cancer controls is 0.017 (grey dashed line). Fig. 12B It was shown that DELFI-TF correlated with the tumor fraction at the time point where the mutant allele fraction of RAS / BRAF assessed by ddPCR was undetectable in the investigated mutant panel as measured by ichorCNA (Spearman correlation, rho=0.54, p<0.001). Fig. 12C Shown are the relative coverage of transcription start site (TSS) positions of a panel of genes highly expressed in colorectal cancer at baseline (brown), during systemic treatment (purple), after metastasis resection (red), and at disease recurrence (black) for patient 65. Changes in trough depth reflect dynamic changes in plasma tumor fraction at longitudinal time points.
[0050] Figures 13A-13G Conclusions: DELFI-TF is a noninvasive biomarker of metastatic disease burden and response to systemic therapy. Fig.13AThe DELFI-TF (blue) and RAS / BRAF MAF (orange) values at baseline showed moderate correlation with the SLD of liver target lesions on pre-treatment imaging scans (DELFI-TF, Spearman correlation, rho = 0.49, p < 0.001; MAF, Spearman correlation, rho = 0.48, p < 0.001). Fig. 13B The results show that the DELFI-TF (blue) and RAS / BRAF MAF (orange) values at baseline showed no significant correlation with the baseline level of carcinoembryonic antigen (DELFI-TF, Spearman correlation, rho=0.10, p=0.427; MAF, Spearman correlation, rho=0.15, p<0.236). Fig. 13C The results showed that the DELFI-TF and MAF values at baseline were significantly lower in patients with a later confirmed partial response (PR) or complete response (CR) obtained by two consecutive RECIST 1.1 measurements (DELFI-TF, Wilcoxon test, p=0.048; MAF, Wilcoxon test, p=0.017). Fig.13D It is shown that the DELFI-TF and MAF values at baseline were significantly lower in patients treated with complete resection (orange) or partial resection (green) surgery after systemic treatment (DELFI-TF, Kruskal-Wallis test, p=0.037; MAF, Kruskal-Wallis test, p=0.011). Fig.13E It is shown that colorectal cancer patients with metachronous metastases (grey) exhibited lower tumor scores assessed by DELFI-TF at baseline than patients presenting with synchronous metastases (green) (Wilcoxon test, p<0.001). Fig.13F Figure 2 shows that in the longitudinal imaging scan, liver metastases are highlighted by blue circles (top). The dynamics of cfDNA tumor fraction (DELFI-TF, MAF) and SLD values (bottom) for study patient 11 are shown. Treatment is represented by the shaded bar. The purple dashed line indicates the time of primary tumor resection in weeks after liver metastasis removal. DELFI-TF and ddPCR MAF for KRAS G12D mutation accurately track the dynamics of disease burden before and after complete resection. Figure 13G It is shown that patients who eventually experienced disease progression (top) more often exhibited increased DELFI-TF values at longitudinal time points compared to patients who never exhibited disease progression (bottom).
[0051] Figures 14A-14D DELFI-TF and clinical characteristics of colorectal cancer are shown. Fig.14AIt was shown that the tumor scores assessed by DELFI-TF and MAF at baseline were equal between colon cancers on the left (brown) or right (beige) side (DELFI-TF, Wilcoxon test, p=0.329; MAF, Wilcoxon test, p=0.515). Fig. 14B Tumor fraction assessed by DELFI-TF at baseline is shown to be equivalent between colon cancers with driver RAS / BRAF mutations (red) or wild type (pink) (Wilcoxon test, p=1). Fig. 14C It shows that the tumor score assessed by DELFI-TF and MAF at baseline is higher in patients who eventually experience disease progression (green) than in patients who never experience disease progression (red) (DELFI-TF, Wilcoxon test, p = 0.03; MAF, Wilcoxon test, p = 0.02). The sum of the maximum diameters at baseline is not different between those who have progressed (green) and those who have never progressed (red) (Wilcoxon test, p = 0.94). Fig.14D Shown is a waterfall plot depicting objective clinical response by change in the sum of longest diameters (SLD). Red, DELFI-TF slope above the median. Cyan, DELFI-TF slope below the median.
[0052] Figures 15A-15B The DELFI-TF dynamics of the study patients are shown. Fig.15A The RAS / BRAF wild-type group is shown. Fig. 15B , RAS / BRAF mutant panel.
[0053] Fig.16 The dynamic changes of DELFI-TF are shown to be related to longitudinal clinical outcomes. The DELFI-TF slope is calculated using all time points available for patients who have a blood draw at least at baseline and within 60 days of disease progression (n=81). On the left, the DELFI-TF slope is colored based on the results of being lower than (cyan) or higher than (red) the median DELFI-TF slope. On the right, a swimming chart, which covers RECIST 1.1, liquid biopsy, surgery and death events of patients sorted in weeks according to the time in the study. Each bar represents the interval between study registration and death or the last follow-up. The bar segments are colored according to RECIST1.1 readings.
[0054] Figures 17A-17D Conclusions: Baseline DELFI-TF and DELFI-TF slope were associated with progression-free survival (PFS) and overall survival (OS). Fig.17AKaplan-Meier curves of PFS according to baseline DELFI-TF values below (orange) or above (blue) the lower quartile among patients with RAS / BRAF mutant and wild-type metastatic colorectal cancer (n=128) are shown (log-rank p=0.015). Fig. 17B Kaplan-Meier curves for PFS according to DELFI-TF slope below (orange) or above (blue) the median among patients with at least one blood draw within 60 days of disease progression are shown (n=81) (log-rank p<0.001). Fig. 17C Kaplan-Meier curves for PFS according to DELFI-TF slope below (orange) or above (blue) the median among patients who experienced an imaging response or stable disease lasting longer than 12 months are shown (n=42) (log-rank p<0.001). Fig.17D Kaplan-Meier curves for OS according to DELFI-TF slope below (orange) or above (blue) the median among patients with at least one blood draw within 60 days of disease progression (n=81) are shown (log-rank p<0.001).
[0055] Figures 18A-18D Imaging and plasma biomarkers showing survival outcomes in patients with metastatic colorectal cancer. Fig.18A Figure 2 Kaplan-Meier curves of progression-free survival (PFS) according to imaging response assessed by RECIST 1.1 show no survival difference between patients with partial response (PR) (orange) or stable disease (SD) (blue). PD, progressive disease. Fig.18B Kaplan-Meier curves of PFS according to baseline MAF below (orange) or above (blue) the lower quartile value of RAS / BRAF among patients with RAS / BRAF mutant metastatic colorectal cancer (n=65) are shown (log-rank p=0.003). Fig. 18C Kaplan-Meier curves for PFS according to baseline carcinoembryonic antigen values below (orange) or above (blue) the lower quartile values among patients with metastatic colorectal cancer (n=127) are shown (log-rank p=0.067). Fig.18D Kaplan-Meier curves of overall survival (OS) according to surgical status and DELFI-TF slope among patients who had at least one blood draw within 60 days of disease progression are shown (n=81) (log-rank p<0.001). DETAILED DESCRIPTION
[0056] A non-invasive method for monitoring cancer in subjects with cancer and predicting overall survival, progression-free survival and the time until progression is described herein. cfDNA in blood can provide a non-invasive way for monitoring the disease of patients with cancer. As demonstrated herein, the whole genome fragmentation pattern of cfDNA in patients with various types of cancer and healthy individuals is evaluated using DNA evaluation of early interception fragments (DELFI). The evaluation of cfDNA includes a scoring method. The defined score (also referred to herein as "DELFI monitoring score") is determined based on the cfDNA fragmentation profile obtained using the cfDNA fragments of a given patient sample. Using the method described herein to evaluate cfDNA can also provide a method for monitoring cancer, and the monitoring can increase the chance of successful treatment and improve the outcome of patients with cancer.
[0057] Before describing the compositions and methods of the present invention, it should be understood that the present invention is not limited to the specific methods and systems described, as such methods and systems may vary. It should also be understood that the terminology used herein is for the purpose of describing specific embodiments only, and is not intended to be limiting, as the scope of the present invention is limited only by the appended claims.
[0058] As used in this specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to "the method" includes one or more methods and / or steps of the type described herein that will become apparent to those skilled in the art upon reading this disclosure and the like.
[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the invention belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, preferred methods and materials are now described.
[0060] The present disclosure provides innovative methods and systems for analyzing cfDNA to monitor, detect or otherwise assess cancer. As indicated by previous studies, on average, individuals without cancer have longer cfDNA fragments (average size of 167.09 bp), while individuals with cancer have shorter cfDNA fragments (average size of 164.88 bp). The method described herein allows a large number of abnormalities in cfDNA to be analyzed simultaneously by performing whole genome analysis of cfDNA fragmentation patterns.
[0061] Thus, in one embodiment, the present invention provides a method for monitoring cancer in a subject. The method comprises:
[0062] determining a cell-free DNA (cfDNA) fragmentation profile of a sample from a subject;
[0063] calculating a score based on the cfDNA fragmentation profile, the score indicating the likelihood of the presence of cancer in the subject;
[0064] determining the ratio of short fragments to long fragments and the fragment size distribution according to the fragmentation profile;
[0065] calculating a divergence score based on the ratio of the short fragments to the long fragments;
[0066] determining a set of model weights based on the fragment size distribution;
[0067] training a machine learning model using a collection of features extracted from a plurality of fragmentation profiles from a plurality of subjects; and
[0068] Based on the monitoring score, the divergence score and the model weight, a monitoring score for the sample is determined by the machine learning model, wherein the monitoring score indicates the level of tumor-derived nucleic acid in the cfDNA of the sample.
[0069] In an embodiment, the present invention provides a method for treating a subject suffering from cancer. The method comprises:
[0070] a) using the method of the invention to detect cancer in the subject, or using the method of the invention to determine the overall survival of the subject; and
[0071] b) administering a cancer therapy to the subject, thereby treating the subject.
[0072] In another embodiment, the present invention provides a method for monitoring cancer in a subject. The method comprises:
[0073] a) using the method of the present invention to detect cancer in the subject, or using the method of the present invention to determine the overall survival of the subject;
[0074] b) administering a cancer treatment to the subject; and
[0075] c) using the method of the invention to determine the overall survival of said subject following administration of said cancer therapy, thereby monitoring said subject's cancer.
[0076] The method described herein utilizes cfDNA fragmentation profile. As used herein, the term "fragmentation profile" (in some aspects, determining the cfDNA fragmentation profile of a mammal) can be used to identify a mammal as having cancer. For example, cfDNA fragments obtained from a mammal (e.g., from a sample obtained from a mammal) can be subjected to low coverage whole genome sequencing, and the sequenced fragments can be mapped to the genome (e.g., in a non-overlapping window) and evaluated to determine the cfDNA fragmentation profile. The cfDNA fragmentation profile of a mammal with cancer is more heterogeneous (e.g., in terms of fragment length) than the cfDNA fragmentation profile of a healthy mammal (e.g., a mammal not suffering from cancer).
[0077] The cfDNA fragmentation atlas can include one or more cfDNA fragmentation patterns. The cfDNA fragmentation pattern can include any appropriate cfDNA fragmentation pattern. Examples of cfDNA fragmentation patterns include, but are not limited to, fragment size density, median fragment size, fragment size distribution, ratio of small cfDNA fragments to large cfDNA fragments, and coverage of cfDNA fragments. In some aspects, the cfDNA fragmentation atlas can be a whole genome cfDNA atlas (e.g., a whole genome cfDNA atlas in a window in the whole genome). In some aspects, the cfDNA fragmentation atlas can be a targeted region atlas. The targeted region can be any appropriate part of the genome (e.g., a chromosome region). Examples of chromosome regions that can be determined as described herein for cfDNA fragmentation atlases include, but are not limited to, a part of chromosomes (e.g., 2q, 4p, 5p, 6q, 7p, 8q, 9q, 10q, 11q, 12q, and / or a part of 14q) and chromosome arms (e.g., 8q, 13q, 11q, and / or 3p chromosome arms). In some cases, a cfDNA fragmentation profile can comprise two or more targeted region profiles.
[0078] In various aspects, the cfDNA obtained from the sample is separated, and a fragment of a specific size range is used in the analysis. In some aspects, the analysis excludes fragment sizes less than about 10bp, 50bp, 100bp or 105bp and greater than about 220bp, 250bp, 300bp, 350bp or greater. In some aspects, the analysis excludes fragment sizes less than 105bp and greater than 170bp. In some aspects, the analysis excludes fragment sizes less than about 230bp, 240bp, 250bp, 260bp and greater than about 420bp, 430bp, 440bp, 450bp or greater. In some aspects, the analysis excludes fragment sizes less than 260bp and greater than 440bp.
[0079] In some aspects, the cfDNA fragmentation profile can be determined by: processing a sample including cfDNA fragments from a subject into a sequencing library; subjecting the sequencing library to low-coverage whole-genome sequencing to obtain sequenced fragments; mapping the sequenced fragments to the genome to obtain a window of mapped sequences; and analyzing the window of mapped sequences to determine the cfDNA fragment length.
[0080] In some aspects, the cfDNA fragmentation profile can be determined by: obtaining cfDNA fragments from a subject and isolating the cfDNA fragments; sequencing the cfDNA fragments to obtain sequenced fragments; mapping the sequenced fragments to the genome to obtain a window of mapped sequences; and analyzing the window of mapped sequences to determine the cfDNA fragment lengths and generate the cfDNA fragmentation profile.
[0081] The method of the present invention is based on low coverage whole genome sequencing and analysis of isolated cfDNA.In one aspect, the data used to develop the method of the present invention is based on shallow whole genome sequence data (1-2x coverage).
[0082] In some aspects, the mapped sequences are analyzed in non-overlapping windows covering the genome. Conceptually, the size of the window can be in the range of thousands to millions of bases, thereby generating hundreds to thousands of windows in the genome. A 5Mb window is used to evaluate the cfDNA fragmentation pattern, because these windows will provide more than 20,000 reads per window, even within a limited amount of 1-2x genome coverage. In each window, the coverage and size distribution of cfDNA fragments are checked. In some aspects, the whole genome pattern from an individual can be compared with the whole genome pattern of a reference population to determine whether the pattern may be healthy or cancer-derived.
[0083] In certain aspects, the mapped sequence comprises tens to thousands of genomic windows, such as 10, 50, 100 to 1,000, 5,000, 10,000 or more windows. Such windows can be non-overlapping or overlapping and can comprise about 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 million base pairs.
[0084] In various aspects, the cfDNA fragmentation profile is determined within each window. Accordingly, the present invention provides methods for determining a cfDNA fragmentation profile of a subject (eg, in a sample obtained from a subject).
[0085] In some aspects, the cfDNA fragmentation profile can be used to identify changes (e.g., alterations) in cfDNA fragment length. The alterations can be changes in the whole genome or in one or more targeted regions / locus. The target region can be any region containing one or more cancer-specific alterations. In some aspects, the cfDNA fragmentation profile can be used to identify (e.g., simultaneously identify) about 10 alterations to about 500 alterations (e.g., about 25 to about 500, about 50 to about 500, about 100 to about 500, about 200 to about 500, about 300 to about 500, about 10 to about 400, about 10 to about 300, about 10 to about 200, about 10 to about 100, about 10 to about 50, about 20 to about 400, about 30 to about 300, about 40 to about 200, about 50 to about 100, about 20 to about 100, about 25 to about 75, about 50 to about 250, or about 100 to about 200 alterations).
[0086] In various aspects, the cfDNA fragmentation atlas may include a cfDNA fragment size pattern. The cfDNA fragments may be of any appropriate size. For example, in some aspects, the length of the cfDNA fragments may be from about 50 base pairs (bp) to about 400bp. As described herein, a subject with cancer may have a cfDNA fragment size pattern containing a median cfDNA fragment size shorter than the median cfDNA fragment size of a healthy subject. A healthy subject (e.g., a subject without cancer) may have a median cfDNA fragment size of about 166.6bp to about 167.2bp (e.g., about 166.9bp) of cfDNA fragment size. In some aspects, a subject with cancer may have an average cfDNA fragment size shorter than the cfDNA fragment size of a healthy subject by about 1.28bp to about 2.49bp (e.g., about 1.88bp). For example, a subject with cancer may have a median cfDNA fragment size of about 164.11bp to about 165.92bp (e.g., about 165.02bp) of cfDNA fragment size.
[0087] In some aspects, the length of the dinucleosome cfDNA fragment can be about 230 base pairs (bp) to about 450bp. As described herein, a subject with cancer can have a dinucleosome cfDNA fragment size pattern shorter than the median dinucleosome cfDNA fragment size of a healthy subject. In some aspects, on average, a subject without cancer has a longer cfDNA fragment (average size of 334.75bp) within the dinucleosome range, while a subject with cancer has a shorter dinucleosome cfDNA fragment (average size of 329.6bp). Thus, a healthy subject (e.g., a subject without cancer) can have a dinucleosome cfDNA fragment size of about 334.75bp with a median cfDNA fragment size. In some aspects, a subject with cancer can have a dinucleosome cfDNA fragment size shorter than the dinucleosome cfDNA fragment size of a healthy subject. For example, a subject having cancer can have a di-nucleosome cfDNA fragment size with a median cfDNA fragment size of approximately 329.6 bp.
[0088] The cfDNA fragmentation atlas can include cfDNA fragment size distribution. As described herein, subjects with cancer can have a more variable cfDNA size distribution than the cfDNA fragment size distribution of healthy subjects. In some aspects, the size distribution can be in the targeted region. Healthy subjects (e.g., subjects without cancer) can have a targeted region cfDNA fragment size distribution of about 1 or less than about 1. In some aspects, subjects with cancer can have a longer (e.g., long 10bp, 15bp, 20bp, 25bp, 30bp, 35bp, 40bp, 45bp, 50bp or larger bp, or any number of base pairs between these numbers) targeted region cfDNA fragment size distribution than the targeted region cfDNA fragment size distribution of healthy subjects. In some aspects, the subject with cancer can have a shorter (e.g., short 10bp, 15bp, 20bp, 25bp, 30bp, 35bp, 40bp, 45bp, 50bp or larger bp, or any number of base pairs between these numbers) target region cfDNA fragment size distribution than the target region cfDNA fragment size distribution of healthy subjects. In some aspects, the subject with cancer can have a target region cfDNA fragment size distribution of less than about 47bp to about 30bp than the target region cfDNA fragment size distribution of healthy subjects. In some aspects, the subject with cancer can have a length of cfDNA fragments that differs by an average of 10bp, 11bp, 12bp, 13bp, 14bp, 15bp, 15bp, 17bp, 18bp, 19bp, 20bp or larger bp. For example, the subject with cancer can have a length of cfDNA fragments that differs by an average of about 13bp target region cfDNA fragment size distribution. In some aspects, size distribution can be a full genome size distribution.
[0089] The cfDNA fragmentation atlas can include the ratio of small cfDNA fragments to large cfDNA fragments and the correlation of fragment ratio with reference to fragment ratio.As used herein, about the ratio of small cfDNA fragments to large cfDNA fragments, small cfDNA fragments can be about 100bp in length to about 150bp in length.As used herein, about the ratio of small cfDNA fragments to large cfDNA fragments, large cfDNA fragments can be about 151bp in length to 220bp in length.As described herein, the fragment ratio correlation of a subject suffering from cancer (e.g., the correlation of cfDNA fragment ratio with reference to DNA fragment ratio (e.g., DNA fragment ratio from one or more healthy subjects)) can be lower than the fragment ratio correlation of healthy subjects (e.g., 1 / 2, 1 / 3, 1 / 4, 1 / 5, 1 / 6, 1 / 7, 1 / 8, 1 / 9, 1 / 10 or less of the fragment ratio correlation of the healthy subjects). The fragment ratio correlation (e.g., correlation of cfDNA fragment ratio with reference DNA fragment ratio (e.g., DNA fragment ratio from one or more healthy subjects)) of healthy subjects (e.g., subjects without cancer) can be about 1 (e.g., about 0.96). In some aspects, the fragment ratio correlation (e.g., correlation of cfDNA fragment ratio with reference DNA fragment ratio (e.g., DNA fragment ratio from one or more healthy subjects)) of subjects with cancer can be about 0.19 to about 0.30 (e.g., about 0.25) lower on average than the fragment ratio correlation (e.g., correlation of cfDNA fragment ratio with reference DNA fragment ratio (e.g., DNA fragment ratio from one or more healthy subjects)) of healthy subjects.
[0090] The method of the present invention further comprises predicting mutant allele fraction (MAF) based on cfDNA fragmentation atlas. The MAF of mutation in DNA is a common value reported by a diagnostic test for a tumor, and represents the score of the DNA molecule containing the mutation concerned analyzed. For the tumor-derived variant identified in the circulating free DNA (cfDNA), MAF represents the score of all cfDNA containing the variant. cfDNA is a combination of tumor-derived DNA and normal cell-derived DNA, and therefore, the MAF value of cloned somatic cell variant is captured as the score of tumor-derived cfDNA. This MAF is related to the circulating tumor DNA score, and therefore can be used as a replacement for the circulating tumor DNA score.
[0091] In various embodiments, the present invention can use the predicted MAF to detect cancer in a subject, predict disease prognosis, predict response to treatment, and / or assess overall survival of a subject. Figure 1It is a block diagram, which shows the process for training a machine learning model to predict MAF. The predicted MAF can be referred to as DELFI monitoring score (DMS). At 102, DELFI scores are calculated based on cfDNA fragmentation patterns. DELFI scores indicate how similar the fragmentation patterns are to the prototype individual with cancer or the individual without cancer. In some aspects, calculating DELFI scores includes: i) determining the ratio of the short cfDNA fragments and long cfDNA fragments of the sample, ii) determining the Z score of the cfDNA fragments of the sample by the chromosome arm, iii) using a calculated mixed model analysis to quantify the cfDNA fragment density, and iv) using a machine learning model to process the output of i)-iii) to define a score. In various aspects, the score is used to determine the possibility of the overall survival of the subject.
[0092] In an illustrative example (Example 1), in a multi-cancer group, the inventors calculated the ratio of short fragments to long fragments calculated by 5MB bins for each individual, the Z score calculated by chromosome arm, and the mixed model of cfDNA fragment size according to low coverage whole genome sequencing. Using these features as input, the inventors fit the cross-validated gradient boosting machine to each person's cancer status (cancer / non-cancer). The output of this model is a score ranging from 0 to 1, where a high value indicates that the signal of cancer is stronger, and a low value indicates that it is more similar to non-cancer. The score generated using these techniques can use features to train a machine learning model to generate a DMS.
[0093] At 104, a DELFI divergence may be calculated. In some aspects, the DELFI divergence may be equal to 1 minus the correlation between the binned and mean-centered short-to-long ratios of a given sample and the binned and mean-centered short-to-long ratios of healthy samples. For example, a healthy sample may be equal to the median of the binned and mean-centered short-to-long ratios of a reference group containing only healthy samples. As used herein, the mean-centered short-to-long ratio is the binned short-to-long ratio minus the overall mean.
[0094] At 106, a set of weights for calculating a hybrid model can be determined. In some aspects, the hybrid model can be a vector containing 11 weights that summarize the fragmentation distribution in the sample. The weights from the hybrid model are estimated using a Bayesian mixed normal distribution of the empirical fragment size distribution.
[0095] At 108, a regression model can be trained for the measured MAF of an individual so that the model learns the features of the DELFI score, DELFI divergence and the mixed model weight of the sample, which contributes to the known MAF of the sample. For example, MAF can be relevant to tumor load, for example, estimated by MAF. In some aspects, the regression model can be a Bayesian hierarchical regression model, which comprises multiple layers, and wherein each layer comprises more predictors. When running, the model uses DELFI score, DELFI divergence and mixed model weight as input, and outputs the predicted MAF. Training is completed by "leaving out one patient" cross validation. In this cross validation scheme, each patient's data is left out successively, and the model is trained on the remaining samples, and then the trained model is used to generate the prediction of the sample left out. In one instance of the model, MAF is a beta-distributed random variable and the model assumes that the expected MAF for a given sample is functionally related to the described features by the inverse logit of the feature matrix multiplied by the vector of regression coefficients plus a patient-specific random intercept that accounts for the intra-patient correlation between the measurements.
[0096] At 110, the trained model can be validated to confirm that it achieves the desired level of accuracy. In various embodiments, the trained model can be evaluated statistically and clinically. For example, the quality of the generated prediction can be evaluated by evaluating the correlation of the predicted tumor burden with the observed tumor burden values. Other examples of validation schemes performed to evaluate the trained model include observing longitudinal graphs that display measured tumor burden values with superimposed predicted tumor burden values and evaluating the relationship between the patient's tumor burden prediction and time to death.
[0097] The model can also be verified in a clinical setting to understand the clinical utility of prediction. In order to verify the model clinically, two metastatic groups that have never been treated were obtained. The group began to use chemotherapy or targeted agents for anti-tumor treatment, and it was determined that the predicted ctDNA load (represented by MAF) could predict the survival of each patient in the group when blood was drawn at baseline and after the first treatment. For 76 patients with metastatic colorectal cancer (mCRC) and 17 patients with metastatic non-small cell lung cancer (mNSCLC), the observed MAF data were obtained. All patient samples were analyzed independently using the DELFI monitoring scoring method. For mCRC patients and mNSCLC patients, blood was drawn after the first treatment, respectively, at 4-12 weeks and 1-3 weeks after treatment. The MAF of clonal variants was measured by digital droplet PCR (RAS / RAF variants) and deep targeted NGS sequencing (EGFR) to mCRC patients and mNSCLC patients, respectively.
[0098] The predicted value of a single threshold value of the modeled ctDNA load is evaluated using the Kaplan-Meier estimator. In some aspects, the threshold value of DMS can be selected for each group by leaving out a patient for cross-validation. In this analysis, a sample is removed, and the threshold value that minimizes the log-rank p value is selected. This process is repeated for each patient in the group, and the median of all optimized threshold values from cross-validation is selected as the final threshold value estimated by Kaplan-Meier. Where available, the Cox proportional hazard model is also used to evaluate the predicted value of the continuous modeling of ctDNA load for progression-free survival and overall survival. In another aspect, other methods for determining a threshold value can be used, such as using a reference set of individuals with no tumor score or a low tumor score.
[0099] Model validation results
[0100] Figure 2 It is a graphical diagram, which shows a comparison diagram between the observed longitudinal MAF data and the predicted tumor burden generated by the model. Specifically, the figure shows the relative MAF track of the mCRC patient obtained from the observed MAF data and the predicted MAF data. Each small figure has drawn the observed longitudinal MAF data with a black line, and has drawn the prediction from the model of superposition with a blue line. The figure also shows the boundaries of 95% Bayesian credible intervals, and the 95% Bayesian credible intervals are superimposed on the black line and the blue line as light blue shaded areas. Each small figure represents the longitudinal data from a subject, and in most cases, the observed track of the MAF of a given patient is closely matched with the predicted MAF (i.e., DMS).
[0101] Figure 3It is a graphical diagram, which shows the comparison between the MAF observed and the predicted MAF without reference to a specific time point or a given sample from which patient. Therefore, the figure contains the DMS and the observed MAF of the entire mCRC group. The solid diagonal line included in the figure represents the absolute fair line between DMS and the observed MAF. As shown in the figure, in most cases, the DELFI monitoring score is quite matched with the observed MAF. In the minority of cases where the prediction is far from the absolute fair line, there is evidence to show that, particularly in the sample with low (i.e. less than 1%) measured MAF, the problem is not the model, but the measurement process. For example, due to tumor and / or metastasis heterogeneity and the clonal evolution that may occur during treatment, the variant of interest may be more difficult to evaluate.
[0102] Figure 4 It is a graphical diagram, which shows the comparison between the observed MAF track of mNSCLC patients and the DMS value from the model trained to the mCRC group. Each small figure has drawn the observed longitudinal MAF data with a black line, and has drawn the prediction from the model of superposition with a blue line. The figure also shows the boundaries of 95% Bayesian credible intervals, and the 95% Bayesian credible intervals are superimposed on the black line and the blue line as light blue shaded areas. Each small figure represents the longitudinal data from an experimenter, and in most cases, the observed track of the MAF of a given patient is closely matched with the predicted MAF (that is, DMS).
[0103] The comparison results show that the model trained by the MAF data from the patient (e.g., mCRC patient) suffering from a type of cancer can be successfully applied to the patient (e.g., mNSCLC group) suffering from different types of cancer. External applicability is the desired feature of the prediction model, because although there are substantial differences (cancer type, sequencing depth, etc.) between the two groups, the quality of prediction is generally high. The external applicability of the prediction model described herein improves the efficiency of prediction model development and training by enabling a prediction model trained on a specific data set to be used for generating useful predictions of patients different from the patients included in the training data set.
[0104] Figure 5It is a graphical diagram, which shows the progression-free survival of a metastatic colorectal cancer group (i.e., mCRC group) in small figure A, and shows the progression-free survival of a metastatic lung cancer group (i.e., mNSCLC group) in small figure B. The figure shows the results from Kaplan-Meier estimation, which uses a group-specific cross-validated threshold to distinguish patients with high DMS and low DMS. In both groups, patients whose DMS at the first time point after treatment was lower than the cross-validated threshold (DELFI monitoring score (-)) showed a longer progression-free survival than patients with high DMS. These results show that DMS can distinguish subgroups of patients who are more likely to survive without progression at the first treatment time point.
[0105] Figure 6 , which shows the total survival of the mCRC group obtained based on DMS in small figure A, and shows the total survival of the mCRC after treatment obtained based on DMS in small figure B. In order to determine the ability of DMS to predict the response to treatment, the total survival data of the mCRC group was obtained. The survival data of each patient is marked with an indicator of whether each patient undergoing surgery undergoes complete resection or incomplete resection. The DMS at the time point before the start of treatment is evaluated for a separate cross-validated distinction threshold. As shown in small figure A, the total survival of patients with DMS below the threshold is longer than the total survival of patients with DMS above the threshold. In addition, in patients undergoing surgery, DMS can further distinguish the total survival of patients undergoing incomplete resection or complete resection in small figure B from patients with a longer predicted total survival of complete resection. In summary, the results illustrate the feasibility of DMS in predicting disease prognosis, even before the start of treatment.
[0106] The MAF of clonal variants is related to ctDNA load, and therefore can be used as a quantitative indicator for estimating the fraction of plasma DNA derived from tumors and the total tumor load of patients. However, during treatment, the gene map of the tumor may change under the selective pressure of treatment. Therefore, for longitudinal measurement of patient response, measuring the MAF of only one variant is limited. In order to evaluate the sensitivity of DMS to changes in tumor DNA during treatment, it was determined whether patients whose MAF of clonal variants in treatment was measured as 0% at the first post-treatment time point would benefit from the analysis performed using DMS.
[0107] Figure 7, which shows the progression-free survival of patients whose clonal variant MAF in the mCRC group was measured as 0% in the first post-treatment assessment in panel A, and shows the DMS obtained by ddPCR of patients whose clonal variant MAF in the mCRC group was measured as 0% in the first post-treatment assessment in panel B relative to MAF. As shown in panel A, 43 patients with 0% MAF in the mCRC group were further separated by DMS relative to progression-free survival. Some of these patients were measured as between 5%-15% DMS, as shown in panel B. This data suggests that the predicted ctDNA load based on the whole genome assessment of the fragment may be more sensitive than conventional MAF.
[0108] In addition, Cox proportional hazard analysis was performed on the mCRC group and the mNSCLC group to evaluate the predictive value of continuous DMS. At both the pre-treatment and first post-treatment time points, DMS was predictive of overall survival in the mCRC group (HR: 19.2, 95% CI: 2.7-138.5, and HR: 400.4, 95% CI: 11.8-13581.0, respectively) and progression-free survival in the mNSCLC group (HR: 67.3, 95% CI: 1.1-4073.6, and HR: 246.5, 95% CI: 2.2-28030.9, respectively).
[0109] Example Hardware Implementation
[0110] Figure 8 An example computer 800 is shown that can be used to implement Figure 1 The training algorithm shown in the figure and generates a DMS value. For example, the computer 800 may include a machine learning system that trains the machine learning model to generate a DMS value as described above or a portion or combination thereof in some embodiments. The computer 800 may be any electronic device that runs a software application derived from a compiled instruction, including but not limited to a personal computer, a server, a smart phone, a media player, an electronic tablet, a game console, an email device, etc. In some embodiments, the computer 800 may include one or more processors 802, one or more input devices 804, one or more display devices 806, one or more network interfaces 808, and one or more computer-readable media 812. Each of these components can be coupled by a bus 810, and in some embodiments, these components can be distributed in multiple physical locations and coupled by a network.
[0111] The display device 806 may be any known display technology, including but not limited to a display device using a liquid crystal display (LCD) or light emitting diode (LED) technology. The processor 802 may use any known processor technology, including but not limited to a graphics processor and a multi-core processor. The input device 804 may be any known input device technology, including but not limited to a keyboard (including a virtual keyboard), a mouse, a trackball, a camera, and a touch-sensitive pad or display. The bus 810 may be any known internal or external bus technology, including but not limited to ISA, EISA, PCI, PCI Express, USB, Serial ATA, or FireWire. The computer-readable medium 812 may be any non-temporary medium that participates in providing instructions to the processor 804 for execution, including but not limited to non-volatile storage media (e.g., optical disks, magnetic disks, flash drives, etc.) or volatile media (e.g., SDRAM, ROM, etc.).
[0112] The computer readable medium 812 may include a computer program for implementing an operating system (e.g., Mac The operating system may be a multi-user, multi-processing, multi-tasking, multi-threaded, real-time operating system, etc. The operating system may perform basic tasks including, but not limited to: recognizing input from input device 804; sending output to display device 806; keeping track of files and directories on computer-readable medium 812; controlling peripheral devices (e.g., disk drives, printers, etc.) that may be controlled directly or through I / O controllers; and managing traffic on bus 810. Network communication instructions 816 may establish and maintain network connections (e.g., software for implementing communication protocols such as TCP / IP, HTTP, Ethernet, telephony, etc.).
[0113] The machine learning instructions 818 may include instructions that enable the computer 800 to function as a machine learning system and / or train a machine learning model to generate DMS values as described herein. The application 820 may be an application that uses or implements the processes and / or other processes described herein. The processes may also be implemented in the operating system 814. For example, the application 820 and / or the operating system may create a task in the application described herein.
[0114] The described features may be implemented in one or more computer programs executable on a programmable system comprising at least one programmable processor coupled to receive data and instructions from and to send data and instructions to a data storage system, at least one input device, and at least one output device. A computer program is a collection of instructions that can be used, directly or indirectly, in a computer to perform some activity or bring about some result. A computer program may be written in any form of programming language, including compiled or interpreted languages (e.g., Objective-C, Java), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.
[0115] For example, a processor suitable for executing a program of instructions may include a general-purpose microprocessor and a special-purpose microprocessor and a sole processor or one of multiple processors or cores of any kind of computer. Typically, a processor may receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer may include a processor for executing instructions and one or more memories for storing instructions and data. Typically, a computer may also include one or more mass storage devices for storing data files, or may be operatively coupled to communicate with the mass storage device; such devices include disks, such as internal hard disks and removable disks; magneto-optical disks; and optical disks. Storage devices suitable for tangibly embodying computer program instructions and data may include all forms of non-volatile memory, for example, semiconductor memory devices (such as EPROM, EEPROM) and flash memory devices; disks, such as internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory may be supplemented by or incorporated into an ASIC (Application Specific Integrated Circuit).
[0116] To provide interaction with a user, the features may be implemented on a computer having a display device (such as an LED or LCD monitor) for displaying information to the user and a keyboard and pointing device (such as a mouse or trackball) through which the user can provide input to the computer.
[0117] Features may be implemented on a computer system comprising back-end components such as a data server, or comprising middleware components such as an application server or an Internet server, or comprising front-end components such as a user computer with a graphical user interface or an Internet browser, or any combination thereof. The components of the system may be connected by any form of digital data communication (such as a communication network) or medium for digital data communication. Examples of communication networks include, for example, telephone networks, LANs, WANs, and computers and networks forming the Internet.
[0118] A computer system may include clients and servers. Clients and servers may generally be remote from each other and may typically interact via a network. The relationship of client and server may arise by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
[0119] One or more features or steps of the disclosed embodiments may be implemented using an API. An API may define one or more parameters that are passed between a calling application and other software code (e.g., an operating system, a library routine, a function) that provides a service, provides data, or performs an operation or calculation.
[0120] An API can be implemented as one or more calls in program code that send or receive one or more parameters through a parameter list or other structure based on a calling convention defined in an API specification document. A parameter can be a constant, a key, a data structure, an object, a class of objects, a variable, a data type, a pointer, an array, a list, or another call. API calls and parameters can be implemented in any programming language. A programming language can define the vocabulary and calling conventions that a programmer will use to access functions that support an API.
[0121] In some embodiments, the API call can report to the application the capabilities of the device running the application, such as input capabilities, output capabilities, processing capabilities, power capabilities, communication capabilities, etc.
[0122] Although various embodiments have been described above, it should be understood that the embodiments are presented by way of example rather than limitation. It will be apparent to those skilled in the relevant art that various changes in form and detail may be made without departing from the spirit and scope. In fact, after reading the above description, it will be apparent to those skilled in the relevant art how to implement alternative embodiments. For example, other steps may be provided, or steps may be eliminated from the described process, and other components may be added to or removed from the described system. Therefore, other embodiments are within the scope of the following claims.
[0123] In addition, it should be understood that any drawings highlighting features and advantages are presented for illustrative purposes only. The disclosed methods and systems are each sufficiently flexible and configurable such that the methods and systems may be utilized in ways other than those shown.
[0124] Although the term "at least one" may be frequently used in the specification, claims, and drawings, the terms "a," "an," "the," "said," etc. also mean "at least one" or "the at least one" in the specification, claims, and drawings.
[0125] Finally, it is Applicant's intent that only claims containing the express language "means for" or "step for" will be interpreted under 35 USC 112(f). Claims that do not expressly contain the phrase "means for" or "step for" will not be interpreted under 35 USC 112(f).
[0126] The methods and systems currently described can be used to detect, predict, treat and / or monitor the cancer status of a subject. Any suitable subject (such as a mammal) can be evaluated, monitored and / or treated as described herein. Examples of some mammals that can be evaluated, monitored and / or treated as described herein include, but are not limited to, humans, primates, such as monkeys, dogs, cats, horses, cattle, pigs, sheep, mice and rats. For example, a person suffering from or suspected of having cancer can be evaluated using the methods described herein, and optionally can be treated with one or more cancer treatments as described herein.
[0127] Subjects suffering from or suspected of having any appropriate type of cancer can monitor, evaluate and / or treat (e.g., by administering one or more cancer treatments to the subject) using the methods and systems described herein. Cancer can be cancer at any stage. In some aspects, cancer can be early stage cancer. In some aspects, cancer can be asymptomatic cancer. In some aspects, cancer can be residual disease and / or recurrence (e.g., after surgical resection and / or after cancer therapy). Cancer can be any type of cancer. Examples of cancer types that can be evaluated, monitored and / or treated as described herein include, but are not limited to, lung cancer, colorectal cancer, prostate cancer, breast cancer, pancreatic cancer, bile duct cancer, liver cancer, CNS cancer, gastric cancer, esophageal cancer, gastrointestinal stromal tumors (GIST), uterine cancer and ovarian cancer. Other types of cancer include, but are not limited to, myeloma, multiple myeloma, B cell lymphoma, follicular lymphoma, lymphocytic leukemia, leukemia and myeloid leukemia. In some aspects, cancer is a solid tumor. In some aspects, cancer is a sarcoma, carcinoma or lymphoma. In some aspects, the cancer is lung cancer, colorectal cancer, prostate cancer, breast cancer, pancreatic cancer, bile duct cancer, liver cancer, CNS cancer, gastric cancer, esophageal cancer, gastrointestinal stromal tumor (GIST), uterine cancer or ovarian cancer. In some aspects, the cancer is a hematological cancer. In some aspects, the cancer is myeloma, multiple myeloma, B cell lymphoma, follicular lymphoma, lymphocytic leukemia, leukemia or myeloid leukemia.
[0128] When treating a subject suffering from or suspected of having cancer as described herein, one or more cancer treatments can be administered to the subject.Cancer treatment can be any appropriate cancer treatment.One or more cancer treatments described herein can be administered to the subject at any appropriate frequency (e.g., once or multiple times in a time period ranging from several days to several weeks).Examples of cancer treatments include, but are not limited to, surgical intervention, adjuvant chemotherapy, neoadjuvant chemotherapy, radiotherapy, hormone therapy, cytotoxic therapy, immunotherapy, adoptive T cell therapy (e.g., chimeric antigen receptors and / or T cells with wild-type or modified T cell receptors), targeted therapy, such as administration of kinase inhibitors (e.g., targeting kinase inhibitors of specific gene lesions such as translocation or mutation) (e.g., kinase inhibitors, antibodies, bispecific antibodies), signal transduction inhibitors, bispecific antibodies or antibody fragments (e.g., BiTE), monoclonal antibodies, immune checkpoint inhibitors, surgery (e.g., surgical resection) or any combination thereof.In some aspects, cancer treatment can reduce the severity of cancer, alleviate the symptoms of cancer and / or reduce the number of cancer cells present in the subject.
[0129] In some aspects, the cancer treatment may be a chemotherapeutic agent. Non-limiting examples of chemotherapeutic agents include amsacrine, azacitidine, axathioprine, bevacizumab (or an antigen-binding fragment thereof), bleomycin, busulfan, carboplatin, capecitabine, chlorambucil, cisplatin, cyclophosphamide, cytarabine, dacarbazine, daunorubicin, docetaxel, doxifluridine, doxorubicin, epirubicin, erlotinib hydrochloride, dapoxetine ... hydrochloride), etoposide, fiudarabine, floxuridine, fludarabine, fluorouracil, gemcitabine, hydroxyurea, idarubicin, ifosfamide, irinotecan, lomustine, mechlorethamine, melphalan, mercaptopurine, methotrexate, mitomycin, mitoxantrone, oxantrone, oxaliplatin, paclitaxel, pemetrexed, procarbazine, all-trans retinoic acid, streptozocin, tafluposide, temozolomide, teniposide, tioguanine, topotecan, uramustine, valrubicin, vinblastine, vincristine, vindesine, vinorelbine, and combinations thereof.Additional examples of anti-cancer therapies are known in the art; see, for example, therapy guidelines from the American Society of Clinical Oncology (ASCO), the European Society for Medical Oncology (ESMO), or the National Comprehensive Cancer Network (NCCN).
[0130] When monitoring a subject having or suspected of having a cancer as described herein, monitoring can be performed before, during and / or after the course of cancer treatment. The monitoring methods provided herein can be used to determine the efficacy of one or more cancer treatments and / or to select subjects for enhanced monitoring.
[0131] In some aspects, monitoring can include conventional techniques capable of monitoring one or more cancer treatments (e.g., the efficacy of one or more cancer treatments). In some aspects, a diagnostic test (e.g., any of the diagnostic tests disclosed herein) can be administered to a subject selected for enhanced monitoring at an increased frequency compared to a subject not selected for enhanced monitoring. For example, a diagnostic test can be administered to a subject selected for enhanced monitoring twice a day, once a day, twice a week, once a week, twice a month, once a month, once a quarter, once every six months, once a year, or any frequency therein.
[0132] In various aspects, DNA is present in a biological sample taken from a subject and is used in the method of the present invention. Biological sample can actually be any type of biological sample comprising DNA. Biological sample is generally a fluid, such as whole blood or a part thereof with circulating cfDNA. In an embodiment, sample comprises DNA from a tumor or liquid biopsy, and the liquid biopsy is such as but not limited to amniotic fluid, aqueous humor, vitreous humor, blood, whole blood, graded blood, plasma, serum, breast milk, cerebrospinal fluid (CSF), cerumen (ear wax), chyle, chyme, endolymph, perilymph, feces, breath, gastric acid, gastric juice, lymph, mucus (including nasal drainage and mucus), pericardial fluid, peritoneal fluid, pleural fluid, pus, nasal mucus, saliva, exhaled breath condensate, sebum, semen, sputum, sweat, synovial fluid, tears, vomitus, prostatic fluid, nipple aspiration fluid, tears, sweat, cheek swab, cell lysate, gastrointestinal fluid, biopsy tissue and urine or other biological fluids. In one aspect, the sample comprises DNA from circulating tumor cells.
[0133] As disclosed above, the biological sample can be a blood sample. The blood sample can be obtained using methods known in the art, such as finger puncture or phlebotomy. Suitably, the blood sample is about 0.1 ml to 20 ml, or alternatively about 1 ml to 15 ml, wherein the volume of blood is about 10 ml. Smaller amounts can also be used, as well as circulating free DNA in the blood. Microsampling and sampling by needle biopsy, catheter, excretion or generation of body fluids containing DNA are also potential sources of biological samples.
[0134] The methods and systems of the present disclosure utilize nucleic acid sequence information and thus may include any method or sequencing apparatus for performing nucleic acid sequencing, including nucleic acid amplification, polymerase chain reaction (PCR), nanopore sequencing, 454 sequencing, insertion tag sequencing. In some aspects, the methods or systems of the present disclosure utilize systems such as those provided by the following companies, as well as other systems: Illumina, Inc (including but not limited to HiSeq TM X10, HiSeq TM 1000, HiSeq TM 2000, HiSeq TM 2500, Genome Analyzers TM 、MiSeq TM , NextSeq, NovaSeq 6000 Systems), Applied Biosystems Life Technologies (SOLiD TM System, IonPGM TM Sequencer, ion Proton TM Sequencer) or Genapsys or BGI MGI. Nucleic acid analysis can also be performed by systems provided by the following companies: Oxford Nanopore Technologies (GridiON TM 、MiniON TM ) or Pacific Biosciences TM RS II or Sequel I or II).
[0135] The present invention includes systems for performing the steps of the disclosed methods and is described in part in terms of functional components and various processing steps. Such functional components and processing steps can be implemented by any number of components, operations, and techniques configured to perform specified functions and achieve various results. For example, the present invention can employ various biological samples, biomarkers, elements, materials, computers, data sources, storage systems and media, information collection techniques and processes, data processing standards, statistical analysis, regression analysis, etc. that can perform various functions.
[0136] Therefore, the present invention further provides a system for monitoring, detecting, analyzing and / or evaluating cancer. In various aspects, the system comprises: (a) a sequencer configured to generate a low coverage whole genome sequencing data set of a sample; and (b) a computer system and / or a processor having functions for performing the method of the present invention.
[0137] In some aspects, the computer system further comprises one or more additional modules. For example, the system may include one or more of an extraction unit and / or a separation unit operable to select a suitable gene component analysis (e.g., cfDNA fragments of a specific size).
[0138] In some aspects, the computer system further comprises a visual display device. The visual display device may be operable to display the curve fit line, the reference curve fit line, and / or a comparison of the two.
[0139] The method for detecting and analyzing according to various aspects of the present invention can be implemented in any suitable manner (e.g., using a computer program operating on a computer system). As discussed herein, the exemplary system according to various aspects of the present invention can be implemented in conjunction with a computer system, which is a conventional computer system such as a processor and a random access memory, such as an application server, a network server, a personal computer, or a workstation that can be accessed remotely. The computer system also suitably includes other memory devices or information storage systems, such as a mass storage system and a user interface, such as a conventional monitor, a keyboard, and a tracking device. However, the computer system can include any suitable computer system and associated equipment, and can be configured in any suitable manner. In one embodiment, the computer system includes an independent system. In another embodiment, the computer system is a part of a computer network including a server and a database.
[0140] Can be implemented in a single device or in multiple devices, implement the software that needs to be used to receive, process and analyze information.Described software can be accessed through a network so that the storage and processing of information occur remotely relative to the user.System according to various aspects of the present invention and its various elements provide functions and operations that are convenient to detect and / or analyze, such as data collection, processing, analysis, reporting and / or diagnosis.For example, in this aspect, a computer system executes a computer program, and described computer program can receive, store, search, analyze and report the information relevant to human genome or its region.Computer program can include multiple modules that perform various functions or operations, such as a processing module for processing raw data and generating supplementary data and an analysis module for analyzing raw data and supplementary data to generate the quantitative assessment of disease state model and / or diagnostic information.
[0141] The programs performed by the system may include any suitable process to facilitate analysis and / or cancer diagnosis. In one embodiment, the system is configured to model a disease state and / or determine a patient's disease state. Determining or identifying a disease state may include generating any useful information about a patient's condition relative to a disease, such as performing a diagnosis, providing information that aids in diagnosis, assessing the stage or progression of a disease, identifying a condition that may indicate susceptibility to a disease, identifying whether further testing may be recommended, predicting and / or assessing the efficacy of one or more treatment procedures, or otherwise assessing a patient's disease state, likelihood of disease, or other health aspects.
[0142] The following example is provided to further illustrate the advantages and features of the present invention, but is not intended to limit the scope of the present invention. Although this example is typical in the examples that can be used, other procedures, methods or techniques known to those skilled in the art may also be used alternatively.
[0143] Example 1
[0144] Detecting cancer and predicting overall survival
[0145] In this example, the methods of the present disclosure were used to detect cancer and predict overall patient survival. Appendix A describes the studies and results.
[0146] This study of prospectively enrolled individuals demonstrated the ability of cfDNA fragmentation assays to distinguish patients with cancer from individuals without cancer. The assays of the present invention exhibit high performance in a multi-cancer setting using only fragmentation-related information obtained from low-coverage WGS.
[0147] The results showed that machine learning models can use cfDNA fragmentation profiles to distinguish cancer from non-cancer despite the presence of common non-malignant conditions (including cardiovascular disease, autoimmune disease, or inflammatory disease). In addition, individuals with higher DELFI scores had a worse prognosis, independent of other characteristics.
[0148] These data support the development of genome-wide cfDNA fragmentation assays for the non-invasive detection of both single and multiple cancers.
[0149] Example 2
[0150] Cell-free DNA fragmentation profiling for monitoring treatment response in metastatic cancer
[0151] The fraction of circulating tumor DNA (ctDNA) molecules in plasma (ctDNA load) has become a feasible measure for describing the total tumor burden of patients with cancer. The ctDNA load can change over time, thereby reducing when the treatment responds and increasing when the tumor is resistant to therapy. Monitoring the dynamics of ctDNA throughout the treatment can enable surgeons to make timely treatment decisions. Ideally, this requires a fast, low-cost and universally applicable monitoring test that predicts treatment success and patient prognosis. Plasma ctDNA from liquid biopsy has great potential as a minimally invasive biomarker for tumor detection and response monitoring to (targeted) therapy. Plasma ctDNA is a dynamic tumor marker produced due to its short half-life, and recurrence can be detected earlier than imaging parameters and clinical parameters.
[0152] There are a variety of techniques for ctDNA profiling. Targeted next-generation sequencing (NGS) is a sensitive method that can provide information about somatic abnormalities and detect genomic changes in tumors. However, this method is limited by the prevalence of clonal hematopoietic variants in the aging population. Tissue or leukocyte-guided methods must be used to prevent these variants from masking the detection of tumor-specific changes. On the other hand, single nucleotide variants can be longitudinally tracked using ctDNA hotspot mutation methods such as droplet digital PCR (ddPCR), which are not too expensive. Because these methods detect a limited number of somatic tumor changes, these hotspot mutation assays are generally not suitable for a variety of tumors within a patient population, and provide a narrow view of the genetic makeup of the tumor. For example, in patients with metastatic colorectal cancer (mCRC), RAS / BRAF driver mutations that can be tracked are present in only half of the patients.
[0153] Previous studies have shown that the size distribution of free DNA (cfDNA) in the genome can be used to reveal its origin. The proportion of shorter fragments is greater in people with cancer than in healthy people. Since the survival of patients with cancer is inversely correlated with the stage of the disease, the cfDNA fragment size composition (i.e., the ratio of shorter cfDNA fragments to longer cfDNA fragments) is used to develop tools for early disease detection in patients with cancer. This method, called DELFI (DNA Evaluation of Early Intercept Fragments), can distinguish cancer from non-cancer and indicate the origin of the tumor. Due to the minimally invasive nature of the technology, cfDNA fragmentomics can also have additional clinical value for monitoring disease progression. Therefore, the DELFI Tumor Fraction Score (DELFI-TF) was developed, which is a machine learning classifier that can detect tumor dynamics without the need for genetic information about the origin tumor. In the work, the DELFI-TF classifier was evaluated for treatment response monitoring in patients with mCRC.
[0154] DELFI-TF model development using genome-wide cfDNA fragmentation profiles - 692 consecutive plasma samples from patients with mCRC and RAS / BRAF mutant (n = 79) or RAS / BRAF wild-type (n = 74) disease participating in a prospective phase III clinical trial (CAIRO5) were processed and analyzed (Table 1, Table 2). Fig.9A and Fig. 9B After confirmation of study eligibility (including unresectable liver metastases as defined by a central expert panel), mutation status of KRAS (exons 2, 3, and 4), NRAS (exons 2 and 3), and BRAF (codon 600) was assessed in available tissue samples followed by blood draws at baseline before treatment and at serial time points during treatment ( Fig.9A During the initial two-month period, patients in both the mutant and wild-type groups were treated with a first-line fluoropyrimidine-based regimen (FOLFOX or FOLFIRI) plus bevacizumab ( Fig.9A and Fig. 10A ). Therefore, radiological evaluations planned for assessing resectability of liver metastases were performed by central tumor board review. Tumor-informed cfDNA analysis was retrospectively performed using droplet digital PCR (ddPCR) in 312 samples from patients in the mutant panel. Tumor-agnostic DELFI-TF analysis was successfully performed in 692 samples from patients in both the mutant and wild-type panels ( Fig. 9B and Fig. 10A ). The DELFI-TF failure rates associated with library preparation and whole genome sequencing (WGS) were 0.42% and 0.29%, respectively ( Fig. 9B ). Participants were planned to be followed up until death or withdrawal from the study.
[0155] Table 1: Demographic and clinical characteristics of the study participants
[0156]
[0157] OS, overall survival; PFS, progression-free survival
[0158] To perform mutation-independent assessment of cancer-specific alterations in cfDNA, we first designed the DELFI-TF model ( Fig. 10B For all cfDNA samples from patients in the mutant panel (n=312), tumor burden was initially quantified as the mutant allele frequency (MAF) of tumor tissue-demonstrated RAS / BRAF variants measured by ddPCR. Whole-genome fragment sequencing statistics (MAF) were obtained by low-coverage WGS of cfDNA libraries using replicate aliquots of cfDNA samples. Fig. 10B A Bayesian hierarchical regression model was trained and cross-validated for MAF of tumor-specific driver RAS / BRAF variants measured by ddPCR in all longitudinal cfDNA samples sequenced in the mutant panel. To generate the predicted DELFI-TF value for each sample, this model took into account the DELFI score from a mixed model using cfDNA fragment size density, the plasma aneuploidy (PA) score, and a weight component ( Fig. 10B Unsupervised cluster analysis was performed using short-to-long ratios of fragment sizes across 504 5-Mb bins and arm-level copy number z scores for 39 chromosome arms across baseline and on-treatment time points in 128 patients with mCRC ( Fig. 10C ). Significantly, fragmentation profile differences were observed across several clinical and demographic characteristics at baseline time points in multiple regions across the entire genome of the vast majority of patients with mCRC, and these differences mostly corresponded to high DELFI-TF values. Similarly, it was observed that most time points associated with progressive disease confirmed by imaging assessment also presented significant heterogeneity and high tumor scores, in contrast to most time points associated with stable disease or radiological response after the start of first-line systemic therapy, which were associated with fewer genomic abnormalities and DELFI-TF values ( Fig. 10C and Fig. 10D These findings suggest that models designed to predict tumor fraction in cfDNA could be used to identify systemic treatment responses in real time in a non-invasive manner.
[0159] DELFI-TF accurately reflects cfDNA mutant allele frequencies and copy number changes - The DELFI-TF model was independently analyzed using non-cancer control samples (n=155) from a cohort of symptomatic patients from Denmark who had previously tested negative for a cancer diagnosis. Compared with untreated samples from the CAIRO5 cohort (n=128), non-cancer control samples exhibited significantly lower DELFI-TF values with an upper limit of the 95% confidence interval (CI) of 0.006. Notably, the DELFI-TF values of untreated samples from patients with mCRC were all significantly higher than 0.006 ( Fig.11A Similar distributions were observed when samples from both cohorts were evaluated with ichorCNA, a tool designed to estimate tumor fraction in ultra-low-pass WGS data ( Fig. 12A). However, non-cancer controls showed a wider range of tumor fraction values, including a higher upper limit of the 95% CI (0.017). In addition, the ichorCNA values of a few CRC samples overlapped with the estimated tumor fraction of non-cancer controls, suggesting that DELFI-TF more accurately reflects the disease burden detected in cfDNA samples.
[0160] Next, the analytical performance of DELFI-TF compared with mutation-based tumor burden assessment was evaluated using ddPCR for RAS / BRAF MAF quantification ( Fig. 11B ). A strong correlation was observed between MAF values and DELFI-TF values (Pearson, r=0.85, p=3.92e-89). Interestingly, a wide range of positive DELFI-TF values was observed in the 60 (20.5%) on-treatment time points where RAS / BRAFMAF was undetectable by ddPCR, suggesting that the DELFI-TF method may be more sensitive for measuring tumor fractions in patients receiving effective anti-tumor therapy. It was further confirmed that these inconsistent time points where RAS / BRAFMAF was undetectable presented DELFI-TF values that were correlated with tumor fractions predicted by ichorCNA (Spearman, rho=0.54, p<0.001) ( Fig. 12B ). The tissue was examined ( Fig. 11C ) samples and plasma samples ( Fig.11D ) were analyzed by low-pass WGS (mean coverage 0.2×) of patient-matched formalin-fixed paraffin-embedded (FFPE) tumor tissue from 104 of the 153 patients. Consistently, aberrant cfDNA fragmentation profiles were found in regions of the genome that were copy-neutral and further affected in regions of copy number variation in tissue samples ( Fig. 11C In addition, we also demonstrated equal analytical performance and significant correlations between tumor fractions assessed by DELFI-TF and ddPCR MAF, with respect to cfDNA copy number of MBD1 (Pearson, DELFI-TF: r = 0.64, p < 0.001; ddPCR MAF: r = -0.67, p < 0.001) and PLGC1 (Pearson, DELFI-TF: r = -0.84, p < 0.001; ddPCR MAF: r = 0.55, p < 0.001), genes commonly deleted and amplified in mCRC, respectively ( Fig.11D). The ability to use relative coverage at the transcription start site (TSS) as a surrogate marker for gene expression in CRC to detect regions of nucleosome depletion was explored. It was observed that for samples at baseline, the relative coverage at the TSS of a panel of approximately 900 genes highly expressed in CRC was significantly lower than that of samples on treatment (Wilcoxon, p<0.001) ( Fig.11E and Fig.11F ), which reflects the dynamic changes detected during disease response and progression ( Fig. 12C ). In conclusion, DELFI-TF accurately captures cancer-specific alterations associated with MAF and copy number aberrations in cfDNA.
[0161] Correlation of DELFI-TF with clinical characteristics and standard imaging assessments - The DELFI-TF method was compared with clinical patient characteristics. A modest correlation was observed between DELFI-TF and ddPCR MAF and the sum of the longest diameters of target metastatic lesions (SLD) in the liver at the untreated time point (DELFI-TF, Spearman, rho = 0.49, p < 0.001; ddPCR MAF, Spearman rho = 0.48, p < 0.001) ( Fig.13A In contrast, no correlation was seen with serum carcinoembryonic antigen (CEA) levels measured before the start of treatment (DELFI-TF, Spearman, rho=0.1, p=0.43; ddPCR MAF, Spearman, ho=0.15, p=0.24) ( Fig. 13B Baseline DELFI-TF and ddPCRMAF tumor scores did correlate with clinical response, as pretreatment levels were significantly lower in patients with later confirmed partial or complete responses than in those with stable disease or progressive disease on two consecutive scans (Wilcoxon, DELFI-TF, p<0.05; ddPCRMAF, p<0.05)( Fig. 13C ). In addition, patients deemed resectable after systemic induction therapy had significantly lower baseline tumor scores (Kruskal-Wallis, DELFI-TF, p<0.05; ddPCR MAF, p<0.05) ( Fig.13D ), as did patients with metachronous disease (Wilcoxon, p < 0.05) ( Fig.13E ). Notably, DELFI-TF and ddPCRMAF tumor scores did not differ by tumor laterality ( Fig.14A ) or RAS / BRAF mutation status ( Fig. 14BAt baseline, DELFI-TF and ddPCRMAF tumor scores were significantly lower in patients who never had disease progression at any time point during the CAIRO5 trial (never progressors) than in patients who had disease progression at some time point during treatment (ever progressors). Fig. 14C On the other hand, analysis using the SLD of target liver lesions at baseline could not distinguish between never-progressors and ever-progressors ( Fig. 14C ). Furthermore, it was demonstrated that DELFI-TF can accurately track longitudinal disease burden dynamics even at late time points in patients treated with curative intent liver metastasis resection ( Fig. 13C Overall, it was demonstrated that former progressors more often showed an increase in DELFI-TF values at early time points and disease progression at later time points compared with never progressors ( Figure 13G ).
[0162] After validating the analytical equivalence between the DELFI-TF model and the ddPCR assay for RAS / BRAF MAF assessment, it was decided to further explore the association between the dynamic changes of DELFI-TF and clinical outcomes. In order to adapt to the longitudinal evolution of continuous DELFI-TF values in a single score, the DELFI-TF slope was calculated, which was defined as the slope of the line that began at the first blood biopsy time point after the start of treatment and ended at the time of disease progression confirmed by RECIST1.1 using linear regression fit to the DELFI-TF value. Then a trend of lower DELFI-TF slopes in patients who experienced partial or complete responses as their best overall responses was observed (Fisher exact test, p=0.1) (Figure 14E). Overall, the temporal analysis of DELFI-TF and ddPCR MAF showed comparable tumor dynamics ( Fig.15A For patients in the wild-type group, only DELFI-TF values could be used for temporal analysis ( Fig. 15B Patients with a DELFI-TF slope below the median had a higher objective radiological response rate to first-line therapy and a longer follow-up duration compared with patients with a DELFI-TF slope above the median ( Fig.16 ).
[0163] Subsequently, baseline DELFI-TF and DELFI-TF slope were associated with survival outcomes. At baseline, patients with DELFI-TF values below the first quartile showed longer median progression-free survival (PFS) than those with DELFI-TF values above the first quartile (13.4 months vs. 10.2 months, hazard ratio [HR] = 1.77, 95% CI 1.12 to 2.78, log-rank p = 0.013) ( Fig.17A For the RAS / BRAF mutant group, baseline MAF for tumor fraction assessment showed a similar difference in median PFS (14.4 months vs. 8.3 months, HR=2.56, 95% CI 1.36 to 4.83, log-rank p=0.00272) ( Fig.18A Baseline serum CEA levels did not predict disease progression or death ( Fig.18B Within the trial, patients were evaluated by an expert panel two months after the start of therapy to assess the resectability of liver metastases. This first clinical response evaluation could not distinguish survival differences between patients with partial response and those with stable disease (11.3 months vs. 11.2 months, HR = 1.13, 95% CI 0.79 to 1.61, log-rank p = 0.52) ( Fig. 18C In the entire study population (13.4 months vs. 10.4 months, HR=2.03, 95% CI 1.247 to 3.318, log-rank p=3.76e-3) ( Fig. 17B ) and in patients who experienced a durable clinical benefit defined as an objective response or stable disease longer than 12 months (16.7 months vs. 13.3 months, HR=2.235, 95% CI 1.097 to 4.553, log-rank p=.023) ( Fig. 17C ), patients with a DELFI-TF slope below the median showed longer PFS. Patients with a DELFI-TF slope below the median also had significantly longer overall survival (OS; 59.4 months vs. 29.1 months, HR=3.05, 95% CI 1.58-5.90, log-rank p=5.135e-4) ( Fig.17D ). It was also observed that survival could be further stratified by resection status, with patients with complete resection of both the primary tumor and liver metastases having better OS outcomes than those with incomplete or no resection ( Fig.17D ).
[0164] Liquid biopsy cfDNA analysis is a new and promising clinical tool in cancer research. The DELFI-TF score, a fragmentomic approach that can quantitatively measure tumor burden, was developed and showed its potential for longitudinal disease monitoring in patients with mCRC.
[0165] Currently, liquid biopsy ctDNA tests for the presence of cancer rely primarily on the detection of one or more somatic tumor alterations. Different research advantages utilize cfDNA fragmentomic features as surrogate features. In vitro and in silico size selection of cfDNA molecules (i.e., selection of shorter cfDNA fragments rather than longer cfDNA fragments) can enrich ctDNA and enhance the identification of genetic alterations in ctDNA. In addition, whole-genome fragmentation maps can facilitate tumor detection and identification of the tumor of origin. The novelty of the cfDNA fragmentomic approach is that low-coverage whole-genome sequencing of trace cfDNA can be used to monitor patient response longitudinally without the need to detect driver mutations.
[0166] Despite advances in diagnosis and treatment, most patients with mCRC will relapse, creating a clinical need for a biomarker to guide the course of treatment. However, currently available follow-up methods, such as clinical imaging and serum CEA, are challenging due to their limited accuracy in detecting tumor tissue viability and assessing treatment effectiveness soon after therapy initiation. The current study suggests that DELFI-TF may be more sensitive than conventional methods for monitoring treatment response, as DELFI-TF is more predictive of PFS than serum CEA measurements and clinical computed tomography (CT) imaging after treatment initiation. Identifying treatment response or progression provides surgeons with the opportunity to adjust their patients' treatment regimens.
[0167] In addition to DELFI-TF, ddPCR MAF after treatment initiation can also predict disease recurrence. However, the ability to detect differences in PFS in patients whose ddPCR MAF is undetectable suggests that DELFI-TF may be more sensitive to treatment response monitoring, but the fragment omics monitoring method cannot track treatment-induced genomic changes in tumors, which is possible in the case of targeted sequencing methods. In addition, both ddPCR MAF and DELFI-TF before treatment indicate the success rate and OS of complete resection of liver metastases. However, compared to hotspot mutation determination (such as ddPCR), DELFI-TF has a more conceptual advantage. Since DELFI-TF does not require prior knowledge of the driver changes of the tumor, it is generally applicable to samples of patients with any type of cancer. The low coverage WGS required for the fragmentation map is lower than the cost of targeted sequencing. Since tumor burden can fluctuate over time, it decreases in treatment response and increases as the tumor becomes resistant to therapy, DELFI-TF can be used as a tool for highlighting the appropriate moment for more detailed set sequencing analysis.
[0168] In a limited number of patients with post-hepatectomy blood samples, postoperative positive DELFI-TF appears to indicate disease recurrence with modest sensitivity. However, pre-surgical blood testing may be a confounding factor in the training cohort. Measurements within 48 hours after surgery showed a spike in DELFI-TF. Since surgery is an invasive procedure, samples extracted too close to the time of surgery may represent wound healing rather than tumor-derived cfDNA. Therefore, the cutoff value of positive DELFI-TF results in samples extracted after complete resection (i.e., minimal residual disease setting) should be further studied.
[0169] Here, DELFI-TF is evaluated, and orthogonal validation is applied to the single nucleotide variant genotyping method of samples collected in well-controlled clinical trials derived from patients with mCRC at the sample level.Thus, DELFI-TF is defined, and its potential prognostic ability to detect disease progression superior to conventional methods for therapeutic response monitoring is shown in the training set.Please note that these results must be confirmed in the validation group, and then other types of cancer or early stage diseases must be evaluated before clinical application.These results cannot be directly transferred to other body fluids such as urine and cerebrospinal fluid, because the distribution of its cfDNA fragments is different.In short, a novel quantitative measurement of ctDNA load has been developed using cfDNA fragment omics.In the training group, DELFI-TF appears to be a non-invasive method for the successful treatment of patients with mCRC that can be used to monitor.
[0170] Study design and population: This study was conducted on participants who participated in the prospective CAIRO5 clinical trial ( NCT02162563) was retrospectively analyzed using liquid biopsies collected from a homogeneous group of patients with mCRC. The Phase III randomized CAIRO5 trial studies the best first-line systemic therapy for patients with histologically proven CRC and isolated previously untreated initially unresectable liver metastases. Patients treated with dual chemotherapy (FOLFOX or FOLFIRI) and bevacizumab and having at least one blood draw before and after treatment were included in this study. All patients were considered unresectable at the time of inclusion, i.e., R0 resection was not possible in a procedure with one surgical intervention. After treatment with dual chemotherapy and bevacizumab, an expert group consisting of liver surgeons and abdominal radiologists evaluated the possibility of local treatment of colorectal liver metastases in patients every two months in accordance with current clinical practice. Clinical follow-up was performed according to the standard of care, which included a clinical review every three months and CT imaging and serum CEA every six months. When liver metastases remain unresectable, chemotherapy is continued in the absence of targeted agents, with a total duration of preoperative and postoperative treatment of six months, and patients are continuously evaluated by serum CEA and CT imaging every two months until disease progression. Follow-up was recorded until September 1, 2021. The trial was approved by the Medical Ethics Committee, conducted in accordance with the Declaration of Helsinki, and patients gave written informed consent for study participation and blood collection for translational research.
[0171] Blood Collection and cfDNA Extraction - Liquid biopsy samples were collected before study treatment (baseline), before surgery, after surgery, and every three months during follow-up until disease progression or completion of treatment. Blood samples were extracted from tubes (Streck, La Vista, USA) and collected centrally at the Netherlands Cancer Institute (Amsterdam, Netherlands). Free plasma was separated by a two-step centrifugation process (1700×g for 10 minutes and 20,000×g for 10 minutes). Free plasma was stored at -80°C until further use. cfDNA was isolated using QIAsymphony (Qiagen, Germany) with an elution volume of 60 μL. The cfDNA concentration was assessed using the QubitTM dsDNA High Sensitivity Assay (ThermoFisher; Waltham, MA, USA). As input for library preparation, aliquots of up to 15 ng were prepared and added to 51 μL using TE buffer when necessary. Aliquots of cfDNA were shipped to the laboratory of Delfi Diagnostics (Baltimore, MD, USA).
[0172] Library preparation and cfDNA sequencing—Upon arrival at the laboratory, the extracted cfDNA was quantified using the TapeStation 4200 (Agilent Technologies; Santa Clara, CA, USA). NGS libraries were constructed as previously described (19) with up to 15 ng of cfDNA input using the NEBNext DNA Library Preparation Kit (New England Biolabs; Ipswich, MA, USA) with four major modifications to the manufacturer's guidelines: 1) the library purification step used the on-bead AMPure XP (Beckman Coulter; Brea, CA, USA) method to minimize sample loss during elution and tube transfer steps; 2) NEBNext end repair, dA tailing, and adapter ligases and buffer volumes were adjusted to accommodate the on-bead AMPure XP strategy when appropriate; 3) Illumina dual-index adapters were used in the ligation reactions; and 4) the cfDNA libraries were amplified with Phusion HotStart Polymerase (Thermo Fisher Scientific; Waltham, MA, USA) for four cycles. WGS library quality was determined using a 2100 Bioanalyzer (Agilent Technologies; Santa Clara, CA, USA) or a TapeStation 4200 (Agilent Technologies; Santa Clara, CA, USA). Next, a total of 96 dual-indexed cfDNA libraries containing samples with different barcodes were pooled together into a single lane of an S4 flow cell and subjected to 100-bp paired-end (200-cycle) WGS sequencing on a NovaSeq 6000 (Illumina; San Diego, CA, USA) with a target of 8× coverage per genome. To limit batch effects, libraries for all samples collected from the same individual were created in the same batch, including duplicate libraries as inter-batch controls and technical replicates of nucleosomal DNA obtained from nuclease-digested human peripheral blood mononuclear cells as intra-batch controls.
[0173] RAS / BRAF mutation analysis—Tumor tissue DNA was analyzed for RAS and BRAF V600E mutations, following routine clinical practice for all patients. For a subset of patients with RAS / BRAF tumor tissue mutations, longitudinal liquid biopsy hotspot mutation analysis was performed by ddPCR (Bio-Rad, Hercules, CA, USA) and fragmentation analysis was performed. ddPCR was used according to the manufacturer's instructions. TM KRAS G12 / G13(#1863506), ddPCR TM KRAS Q61(#12001626), ddPCR TM KRAS A146T (#10049550) and ddPCR TM BRAF V600 (#12001037) screening kit, using 9 μL sample, 11 μL ddPCR probe method supermix (without dUTP), 1 μL multiplex assay, and 1 μL nuclease-free water. All measurements were performed in duplicate, including blanks (nuclease-free water) and positive controls. Patients with RAS / BRAF mutations that could not be followed by ddPCR were excluded ( Fig. 9B ). Using QuantaSoft TM Data were analyzed using BioRad software version 1.6.6 (Bio-Rad, Hercules, CA, USA) and the automatic correction algorithm described previously.
[0174] cfDNA sequencing data analysis-On a per-sample basis, paired-end sequencing reads were aligned to the reference genome (hg19) using paired-end alignment using Bowtie (version 2.3.0). The aligned reads were sorted and converted to BAM format using Samtools (version 1.3.1) and Bedtools (version 2.26.0), and then converted to BED format. The fragment lengths were calculated based on the start and end coordinates, and the fragments were divided into 504 5-Mb bins, covering approximately 2.6Gb of the genome. Next, R / Bioconductor (version 3.6.2) was used to calculate the number of short (100-150bp) and long (151-220bp) fragments for each bin, and these counts were corrected by GC content as described by Benjamini and Speed. By bin, the counts of the corrected short fragments were divided by the counts of the corrected long fragments to obtain a fragmentation map for each person.
[0175] Four statistical values for each sample are calculated to generate DELFI-TF scores, DELFI scores, DELFI divergences, mixed model components, and arm-level aneuploidy scores. Similar to the method described by Cristiano et al. (Genome-wide cell-free DNA fragmentation in patients with cancer). "Nature", 570, 385-389 (2019)) DELFI scores are calculated, and the scores indicate how similar the fragmentation profiles look to individuals with cancer or individuals without cancer. DELFI divergences are defined as the correlation between the binning and mean-centered short-long ratio of a given sample minus the binning and mean-centered short-long ratio of a reference group containing only healthy samples. The mixed model summarizes the fragment size distribution, and the weight statistics from this model are evaluated when generating DELFI-TF.
[0176] Using these statistical values calculated for each sample, the Bayesian hierarchical regression model is trained using R for the allele frequency of tumor-specific RAS / BRAF variants measured by ddPCR in longitudinal cfDNA samples. This model uses DELFI scores, DELFI divergences, mixed model weights and aneuploidy scores as input and outputs the predicted MAF. The model assumes that MAF is a random variable of beta distribution, and assumes that the expected MAF of a given sample is functionally related to the described features by the inverse logit of the feature matrix multiplied by the vector of regression coefficients plus the patient-specific random intercept, which illustrates the intra-patient correlation between the measurement results. In order to generate unbiased predictions, avoid overfitting, and evaluate universality, training is completed by leaving out a patient cross-validation. In this cross-validation scheme, the data of each patient is left out in turn, the model is trained on the remaining samples, and then the trained model is used to generate the prediction of the sample left out. DELFI-TF is defined as the predicted MAF generated by the cross-validation scheme. The quality of the generated predictions was evaluated by assessing their correlation with observed ddPCR MAF values and evaluating the relationship between these predictions and time until progression or death.
[0177] DELFI-TF dynamic analysis - To capture the molecular dynamics of tumor burden over time, the DELFI-TF slope was calculated, i.e., the slope of the regression line fitted to the DELFI-TF values within 60 days after the progression date of the PFS analysis and until the progression date of the OS analysis from time T1 onwards. For this practice, patients with at least 3 collected samples before progression were selected, and at least one of those samples was collected in the progression window, which was 120 days before the progression date of the PFS analysis (79 patients) and 120 days before the progression date of the OS analysis until 60 days after the progression (80 patients). The regression line was calculated using Python / scikit-learn (version 3.9.13 / 1.1.1).
[0178] Relative coverage calculation for gene expression analysis - For this analysis, a collection of 854 transcripts identified by the Broad GDACFirehose pipeline, known to be highly expressed in colon adenocarcinoma, was selected, and their transcription start site (TSS) coordinates were extracted. In only 126 patients with plasma samples at T0 and T1 time points, the fragment coverage of these TSSs plus the flanking regions of 1,500 bp on each side for all genes was calculated. The list of TSS coordinates and aligned fragments was in BED format, and coverage was calculated using pybedtools (version 0.9.0), i.e., the python interface to Bedtools.
[0179] Statistical analysis-The correlation between DELFI-TF and ddPCR MAF was calculated using the Pearson correlation coefficient. Similarly, the correlation between DELFI-TF / ddPCR MAF and the copy number ratio was calculated using the Pearson correlation. The Spearman correlation test was used between DELFI-TF / ddPCR MAF and SLD and serum CEA. Except for the survival analysis, all two-sample hypothesis tests used the Wilcoxon rank sum test. The Kruskal-Wallis test was used to compare the tumor scores based on DELFI-TF and MAF between the resection status. The Mantel-Cox log-rank test was used for survival analysis. The analysis was performed using R statistical software (version 4.2.1, Foundation for Statistical Computing, Vienna, Austria). Unless otherwise stated, hypothesis tests were two-sided with a type 1 error of 5% for determining statistical significance.
[0180] Although the present invention has been described with reference to the presently preferred embodiments, it should be understood that various modifications can be made without departing from the spirit of the invention.Accordingly, the present invention is limited only by the following claims.
Claims
1. A method for monitoring cancer, the method comprising: determining a cell-free DNA (cfDNA) fragmentation profile of a sample from a subject; Calculating a fragmentation score based on the cfDNA fragmentation profile, wherein the score indicates the likelihood of the presence of cancer in the subject; determining the ratio of short fragments to long fragments and the fragment size distribution according to the fragmentation profile; calculating a divergence score based on the ratio of the short fragments to the long fragments in the sample as related to the ratio in a sample from a healthy subject; determining a set of model weights based on the fragment size distribution; training a machine learning model using a collection of features extracted from multiple fragmentation profiles of multiple subjects; as well as Based on the fragmentation score, the divergence score and the model weight, a monitoring score for the sample is determined by the machine learning model, wherein the monitoring score indicates the level of tumor-derived nucleic acid in the cfDNA of the sample.
2. The method of claim 1, wherein the divergence score indicates a correlation between the fragmentation pattern and a median fragmentation pattern observed for healthy subjects. The method according to claim 1 , wherein the monitoring score is in the range of 0 to 1.
4. The method of claim 1, further comprising determining at least one of the likelihood of overall survival, the likelihood of progression-free survival, or the time until progression of the subject based on the monitoring score.
5. The method of claim 4, wherein the likelihood of the overall survival, the likelihood of the progression-free survival, or the time until progression of the subject decreases with increasing monitoring score values.
6. The method of claim 5, further comprising classifying the monitoring score as a high score or a low score, wherein a high score indicates at least one of a decreased overall survival, a decreased progression-free survival, or a decreased time until progression for the subject.
7. The method of claim 6, wherein the monitoring score is classified prior to administering a cancer treatment to the subject to provide a baseline classification for the subject.
8. The method of claim 6, wherein the monitoring score is classified at a first time point after a cancer treatment is administered to the subject to provide a post-treatment classification of the subject.
9. The method according to claim 1, wherein the likelihood of cancer progression in the subject increases with an increasing monitoring score value.
10. The method of claim 1, wherein the likelihood of cancer progression in the subject decreases as the monitoring score value decreases.
11. The method of claim 1, wherein the likelihood of a positive response to a cancer treatment administered to the subject increases as the monitoring score value decreases.
12. The method of claim 1, wherein the likelihood of a positive response to a cancer treatment administered to the subject decreases with increasing monitoring score values.
13. The method of claim 1, wherein the monitoring score predicts the fraction of tumor-derived nucleic acid or the fraction of clonal mutant alleles of a mutation associated with cancer.
14. The method of claim 1, wherein the cancer is a solid tumor.
15. The method of claim 1, wherein the cancer is a sarcoma, carcinoma, or lymphoma.
16. The method of claim 1, wherein the cancer is selected from the group consisting of colorectal cancer, lung cancer, kidney cancer, brain cancer, prostate cancer, breast cancer, pancreatic cancer, bile duct cancer, liver cancer, CNS cancer, gastric cancer, esophageal cancer, gastrointestinal stromal tumor (GIST), uterine cancer, and ovarian cancer.
17. The method of claim 1, wherein the cancer is a hematological cancer.
18. The method of claim 1, wherein the cancer is selected from the group consisting of myeloma, multiple myeloma, B-cell lymphoma, follicular lymphoma, lymphocytic leukemia, leukemia, and myeloid leukemia.
19. The method of claim 1, further comprising administering a cancer therapy to the subject.
20. The method of claim 17, wherein the cancer treatment is selected from the group consisting of surgery, adjuvant chemotherapy, neoadjuvant chemotherapy, radiation therapy, hormone therapy, cytotoxic therapy, immunotherapy, adoptive T cell therapy, targeted therapy, or any combination thereof.
21. The method of claim 1, wherein the cfDNA fragmentation profile is determined by: obtaining cfDNA fragments from the subject and isolating the cfDNA fragments; sequencing the cfDNA fragments to obtain sequenced fragments; mapping the sequenced fragments to a genome to obtain a window of mapped sequences; and The windows of mapped sequences are analyzed to determine cfDNA fragment lengths and generate the cfDNA fragmentation profile.
22. A method of determining at least one of overall survival, progression-free survival, or time until progression, the method comprising: determining a cell-free DNA (cfDNA) fragmentation profile of a sample from a subject; Calculating a fragmentation score based on the cfDNA fragmentation profile, wherein the score indicates the likelihood of the presence of cancer in the subject; determining, from the fragmentation profile in the sample, a ratio of short to long fragments and a fragment size distribution as correlated to ratios in a sample from a healthy subject; calculating a divergence score based on the ratio of the short fragments to the long fragments; determining a set of model weights based on the fragment size distribution; Training a machine learning model using a collection of features extracted from fragmented atlases of multiple subjects; Based on the fragmentation score, the divergence score and the model weight, determining a monitoring score for the sample by the machine learning model, wherein the monitoring score indicates the tumorigenic level in the cfDNA of the sample, thereby indicating the possibility of cancer progression in the subject; as well as At least one of overall survival, progression-free survival, or time until progression is determined based on the monitoring score.
23. A system for monitoring cancer in a subject, the system comprising: Memory; as well as One or more processors coupled to the memory, the one or more processors configured to execute operations that cause the computer system to: determining a cell-free DNA (cfDNA) fragmentation profile of a sample from the subject; Calculating a fragmentation score based on the cfDNA fragmentation profile, wherein the score indicates the likelihood of the presence of cancer in the subject; determining, from the fragmentation profile in the sample, a ratio of short to long fragments and a fragment size distribution as correlated to ratios in a sample from a healthy subject; calculating a divergence score based on the ratio of the short fragments to the long fragments; determining a set of model weights based on the fragment size distribution; Training a machine learning model using a collection of features extracted from fragmented atlases of multiple subjects; as well as Based on the fragmentation score, the divergence score and the model weight, a monitoring score for the sample is determined by the machine learning model, wherein the monitoring score indicates a tumor-derived level in the cfDNA of the sample, wherein the tumor-derived level in the cfDNA indicates a possibility of cancer progression in the subject.
Citation Information
Cited By
Establishment method and application of cfDNA fragment map of early-stage bladder cancer
CN119580850A
