Methods for predicting treatment response in cancer

Abstract

The present invention provides a method for predicting the therapeutic response of a cancer patient using a tumor copy number profile for the patient. The method includes analyzing the copy number profile to evaluate whether a characteristic of at least one copy number feature indicates the presence of focal amplification in the tumor genome, the at least one copy number feature being selected from copy number change points, segment size, and segment copy number. If the characteristic of the at least one copy number feature indicates the presence of focal amplification in the tumor genome, the patient is predicted to be likely to be resistant to treatment with an agent that induces micronucleus formation (e.g., doxorubicin). Related methods and systems are also provided.

Description

[Technical Field]

[0001] The present invention relates, in part, to a method for predicting cancer response to chemotherapy based on copy number variation characteristics, particularly in cancers with high chromosomal instability and with respect to chemotherapy that induces micronucleus formation, such as doxorubicin. [Background technology]

[0002] Lung, ovarian, esophageal, sarcoma, glioblastoma, and pancreatic cancers are highly lethal cancers, with minimal improvement in survival rates over the past 40 years. These cancers are characterized by extreme chromosomal instability, resulting in several "actionable" hits from gene panel-based testing. However, due to their unstable genomes, these cancers often have low-frequency recurrent oncogenic mutations, few recurrent copy number variations, and highly complex genomic profiles. Consequently, hits from gene panel-based testing are frequently false positives, and targeted therapies often fail to demonstrate efficacy.

[0003] In the case of ovarian cancer, current genomic stratification is limited to defining homologous recombination repair deficiency (HRD) tumors, with approximately 20% of high-grade serous ovarian cancers (HGSOCs) having BRCA1 / 2 mutations. The standard of care for ovarian cancer is typically a combination of chemotherapy (with carboplatin and paclitaxel) and tumor debulking surgery. Patients who relapse within six months are considered platinum-resistant, while those who do not are considered platinum-sensitive. However, 80% of all patients will experience a gradual return of the disease. Therefore, platinum-resistant patients will receive paclitaxel as monotherapy, followed by doxorubicin if they do not respond to paclitaxel. Platinum-sensitive patients will go through carboplatin again, but this time in combination with doxorubicin. This means that almost all patients will receive doxorubicin at some point. However, the response rate to doxorubicin is low (18% in platinum-resistant patients and 52% in platinum-sensitive patients; Mutch et al., 2012; Kaye et al., 2012; Ferrandina et al., 2008; Gordon et al., Puiade-Lauraine et al., 2012; Rose et al., 2007; Bafaloukos et al., 2010; Alberts et al., 2008; O'Byrne et al., 2002). The evidence suggests that patients with BRCA1 / 2 mutants are slightly more likely to respond, but the effect is small (Kaye et al., 2012).

[0004] While cancer genome analysis holds promise in the field of cancer care, there remains an unmet need for methods to predict treatment response in patients with cancers exhibiting high chromosomal instability. This method aims to provide a solution to these needs and offers further relevant benefits. [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] The inventors have previously demonstrated that copy number signatures representing genomic imprints of obvious mutation processes can be identified in cancers with high chromosomal instability, such as HGSOC. The inventors further discovered that exposure to copy number signatures in diagnosis is a predictor of overall survival, and that exposure to signature 1 is a predictor of platinum-resistant recurrence in HGSOC. Here, the inventors hypothesized that one of these signatures, characterized by copy number features thought to indicate lesion amplification, is associated with the presence of micronuclei during tumor progression. The inventors further hypothesized that the presence of this signature within a tumor may indicate that the tumor is able to tolerate micronucleus formation and, as such, may be resistant to genotoxic chemotherapy that induces micronucleus formation, such as DNA insertors, and particularly DNA insertors acting as topoisomerase II toxins. The inventors have discovered that the presence of a genome-wide copy number feature indicating lesion amplification is associated with resistance to doxorubicin, a DNA insertion agent that acts as a topoisomerase II toxin and induces micronucleus formation in ovarian cancer. This supports a presumptive relationship between the signature, the presence of micronuclei, and resistance to them, and therefore, the presence of this signature can be used to predict resistance to micronucleus-inducing drugs. [Means for solving the problem]

[0006] Therefore, in the first embodiment, the present invention is a method for predicting the treatment response of cancer patients, (a) A step of obtaining a tumor copy number profile for the patient, (b) The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, At least one copy number feature is selected from the copy number change point, segment size, and copy number. The method provides a way to predict that a patient is likely to be resistant to treatment with drugs that induce micronucleus formation if the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome.

[0007] Patients are preferably those diagnosed with or likely to have cancer with high chromosomal instability. Patients may also be diagnosed with epithelial malignancies or sarcomas. Patients may also be diagnosed with glioblastoma, lung cancer, esophageal cancer, pancreatic cancer, breast cancer, or ovarian cancer. Patients are preferably those with high-grade serous ovarian cancer (HGSOC) or triple-negative breast cancer. Tumor copy number profiles for patients may be obtained from tumor samples from the patient or from liquid biopsy samples from the patient, such as blood or plasma samples.

[0008] The agent that induces micronucleus formation may be a genotoxic chemotherapeutic agent. The agent that induces micronucleus formation may be a genotoxic chemotherapeutic agent that acts as a DNA insertion agent and / or a topoisomerase II toxin. The agent is preferably an anthracycline. The agent may be doxorubicin. The agent may be liposomal doxorubicin. Doxorubicin may be pegylated doxorubicin or non-pegylated doxorubicin.

[0009] The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, i. For each copy count event in the copy count profile, quantify one or more copy count features selected from the segment copy count, copy count change point, and segment size, ii. This includes obtaining one or more aggregation measures for each quantified copy number feature.

[0010] The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, iii. For each copy count event in the copy count profile, quantify the copy count change point, iv. Further includes obtaining one or more aggregation measures for copy number change points.

[0011] The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, i. For each copy number event in the copy number profile, one or more further copy number features selected from the number of breakpoints per xMb, the number of breakpoints per chromosome arm, and the length of the segment having the vibrational copy number, where x is preferably 10. ii. This may include obtaining one or more aggregation measures for each further quantified copy number feature.

[0012] In embodiments, one or more aggregation measures include a statistical measure of the centrality, preferably the mean or median, of the distribution of values ​​across copy number events, for copy number features selected from segment copy number and copy number change points. In embodiments, one or more aggregation measures include the number or proportion of values ​​across copy number events that are above and / or below a default threshold. In embodiments, one or more aggregation measures include the maximum value across copy number events, or the maximum value such that the proportion of copy number events at or above that value exceeds a default threshold. In embodiments, one or more aggregation measures include the minimum value across copy number events, or the minimum value such that the proportion of copy number events at or below that value exceeds a default threshold, for segment size copy number features. In embodiments, one or more aggregation measures include the sum of the posterior probabilities of copy number feature values ​​for each copy number event belonging to one or more default distributions.

[0013] In an embodiment, the step of analyzing the copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome includes comparing the mean, median, or maximum segment copy number to a predetermined threshold, wherein a mean, median, or maximum segment copy number exceeding the predetermined threshold indicates the presence of lesion amplification in the tumor genome, and preferably the predetermined threshold is determined by comparing the mean, median, or maximum segment copy number between resistant and susceptible samples in a training group. In an embodiment, the step of analyzing the copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome includes comparing the mean, median, or maximum copy number change point to a predetermined threshold, wherein a mean, median, or maximum copy number change point exceeding the predetermined threshold indicates the presence of lesion amplification in the tumor genome, and preferably the predetermined threshold is determined by comparing the mean or median copy number change point between resistant and susceptible samples in a training group.

[0014] In the embodiment, the default threshold is determined by comparing the mean, median, or maximum copy number change point (or segment copy number) between resistant and susceptible samples in the training group and identifying a threshold that results in the highest proportion of samples being correctly identified as resistant and susceptible (i.e., a threshold that maximizes the accuracy of the classification using the threshold).

[0015] In embodiments, the default threshold is determined by comparing the mean, median, or maximum copy number change point (or segment copy number) between resistant and susceptible samples in a training group to identify a threshold that results in the highest proportion of resistant samples being correctly identified as resistant, while no susceptible samples are misidentified as resistant (i.e., a threshold that maximizes the specificity of the classification using the threshold—the true negative rate—while having a 100% susceptibility—true positive rate).

[0016] In embodiments, the default threshold for the average segment copy number is 2.65–3.60 (or 2.8–3.4, 3–3.2, or about 3.15, e.g., 3.146530). In embodiments, the default thresholds for the median segment copy number are 2.85 and 3.1 (or 2.9–3.05, 2.95–3.05, or about 3, e.g., 3.000824). In embodiments, the default threshold for the maximum segment copy number is 7–9 (or 7.5–8.5, 8–9, or about 8.5). In embodiments, the default threshold for the average copy number change point is 1.2–1.328 (or 1.2–1.328, 1.3–1.328, or about 1.328, e.g., 1.323483). In an embodiment, the default threshold for the median copy number change point is 1.04–2.1, 1.2–2.1, or 1.04–1.06 (e.g., 1.051904). In an embodiment, the default threshold for the maximum copy number change point is 4.01–30, 4.01–5 (or 4.01–4.025), or 17.6–18 (e.g., about 18), or 18–30.

[0017] In an embodiment, the step of analyzing a copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome includes comparing the number or proportion of events having segment copy numbers above a first default threshold to a second default threshold, where the number or proportion of events exceeding the second default threshold indicates the presence of lesion amplification in the tumor genome, and preferably the first and / or second default thresholds are obtained by comparing the distribution of segment copy numbers for resistant and susceptible samples within a training group.

[0018] In an embodiment, the step of analyzing a copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of focal amplifications in a tumor genome comprises comparing (a) the proportion or number of events having a segment size exceeding a first predetermined threshold with a second predetermined threshold, and (b) the proportion or number of events having a segment size below a third predetermined threshold with a fourth predetermined threshold, wherein the proportions in (a) and (b) exceeding their respective predetermined thresholds indicates the presence of focal amplifications in the tumor genome, or comparing the sum of the proportion or number of events having a segment size exceeding a first predetermined threshold and the proportion or number of events having a segment size below a third predetermined threshold with a fifth predetermined threshold, preferably, the first, second, third, fourth, and / or fifth predetermined threshold is obtained by comparing the distribution of segment sizes for resistant and sensitive samples within a training cohort, the comparing.

[0019] In an embodiment, a predetermined threshold is determined by comparing the distribution of segment sizes, copy number change points, and / or segment copy numbers between resistant and sensitive samples within a training cohort and identifying a threshold that results in the highest proportion of samples being correctly identified as resistant and sensitive (i.e., a threshold that maximizes the accuracy of classification using the above threshold).

[0020] In an embodiment, a predetermined threshold is determined by comparing the distribution of segment sizes, copy number change points, and / or segment copy numbers between resistant and sensitive samples within a training cohort and identifying a threshold that results in the highest proportion of resistant samples being correctly identified as resistant while no sensitive samples are incorrectly identified as resistant (i.e., a threshold that maximizes the specificity - true negative rate - of classification using the above threshold while having 100% sensitivity - true positive rate).

[0021] In an embodiment, the first default threshold for the number of segment copies is 5, and the second default threshold is 36-60 (or 36-50, 38-48, 40-45, e.g., 42) (for the number of events) and about 5% (or 25-28%) (for the percentage of events). In an embodiment, the first default threshold for the copy number change point is 4, and the second default threshold is 2-18 (or 2-8, 2-6, e.g., 3) (for the number of events) and 2%-10% (or 2-8%, e.g., 5%) or 8.6-10%, e.g., 8.6%) (for the percentage of events). In an embodiment, the first default threshold for the segment size is 12,000,000 bp, and the third default threshold for the segment size is 4,000,000 bp. In some such embodiments, a second default threshold for segment size is 52–68, for example, 52 or 68 (for the number of events). In some embodiments, a fourth default threshold is 55–214, for example, 55 or 214 (for the number of events). In embodiments, a fifth default threshold is 130 (for the total number of events).

[0022] Advantageously, copy number features are characterized by genome-wide criteria. In one embodiment, the step of analyzing a copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome includes, for each quantified copy number feature, comparing the sum of the posterior probabilities of the copy number feature values ​​for each copy number event belonging to one or more default distributions to the respective default threshold, One or more default distributions are: • Distribution of copy number change point values ​​(C1) centered around values ​​of 2 to 32, preferably 4 to 30, 5 to 30, or 6 to 30. · A distribution of copy values ​​(C2) centered around the values ​​of 5-34, preferably 6-34, 7-34, 6-32, 7-32, or 8-32, and / or · 100,000 to 4,000,000 base pairs, preferably 200,000 to 4,000,000 bp, 300,000 to 4,000,000 bp, 400,000 to 4,000,000 bp, 100,000 to 3,000,000 bp, 200,000 to 3,000,000 bp, 300,000 to 3,000,000 bp, 400,000 to 3,000,000 bp, 100,000 to 2,5 This includes a distribution of segment sizes (C3) centered around values ​​of 00,000 bp, 200,000 to 2,500,000 bp, 300,000 to 2,500,000 bp, or 400,000 to 2,500,000 bp, and a distribution of segment sizes (C4) centered around values ​​of 12,000,000 to 80,000,000 bp, preferably 15,000,000 to 60,000,000. Each default threshold is quantified by comparing the sum of posterior probabilities for resistant and susceptible samples within the training group.

[0023] Distribution C1 is preferably a Gaussian distribution such as a Gaussian distribution with a mean of about 28.7 (e.g., 28.7 ± 10%) and a standard deviation of about 22.1 (e.g., 22.1 ± 10%) (e.g., copy number change point component 7, cp7, etc. in Table 1), a Gaussian distribution with a mean of about 7.3 (e.g., 7.3 ± 10%) and a standard deviation of about 3.4 (e.g., 3.4 ± 10%) (e.g., copy number change point component 6, cp6, etc. in Table 1), or a Gaussian distribution with a mean of about 3.0 (e.g., 3.0 ± 10%) and a standard deviation of about 1.0 (e.g., 1.0 ± 10%) (e.g., copy number change point component 5, cp5, etc. in Table 1).

[0024] In one embodiment, the default threshold for the sum of the posterior probabilities of copy number change points is 0.02041483.

[0025] In one embodiment, distribution C1 is a Gaussian distribution such as copy number change point component 7, cp7, etc. in Table 1, with a mean of approximately 28.7 (e.g., 28.7 ± 10%) and a standard deviation of approximately 22.1 (e.g., 22.1 ± 10%), with default thresholds of 1.91e-5 to 1.23e-3, 1.91e-5 to 1.1e-3, or 1.035e-3 to 1.23e-3. In another embodiment, distribution C1 is a Gaussian distribution such as copy number change point component 6, cp6, etc. in Table 1, with a mean of approximately 7.3 (e.g., 7.3 ± 10%) and a standard deviation of approximately 3.4 (e.g., 3.4 ± 10%), with default thresholds of 2.8e-4 to 6.6e-3, 2.8e-4 to 6.6e-3, or 4.72e-3 to 6.6e-3. In the embodiment, the distribution C1 is a Gaussian distribution with a mean of approximately 3.0 (e.g., 3.0 ± 10%) and a standard deviation of approximately 1.0 (e.g., copy number change point component 5, cp5, etc. in Table 1), and the default threshold is 0.006 to 0.023.

[0026] Distribution C2 is preferably a Gaussian distribution such as copy number component 8, cn8 in Table 1, with a mean of about 30.9 (e.g., ±10%) and a standard deviation of about 23.1 (e.g., ±10%), or a Gaussian distribution such as copy number component 7, cn7 in Table 1, with a mean of about 8.4 (e.g., ±10%) and a standard deviation of 3.5 (e.g., ±10%). In the embodiment, the default threshold for the sum of the segment copy number posterior probabilities is 2e-3 to 1.5e-2 (e.g., 0.006882869). In one embodiment, distribution C2 is a Gaussian distribution such as a Gaussian distribution with a mean of approximately 30.9 (e.g., ±10%) and a standard deviation of approximately 23.1 (e.g., copy number component 8, cn8, etc. in Table 1), with default thresholds of 7e-5 to 6e-3, 7e-5 to 1.1e-4, approximately 1.5e-3, or 2e-3 to 6e-3. In another embodiment, distribution C2 is a Gaussian distribution with a mean of approximately 8.4 (e.g., ±10%) and a standard deviation of 3.5 (e.g., copy number component 7, cn7, etc. in Table 1), with default thresholds of 0.0017 to 0.017, approximately 0.0017, or 0.012 to 0.017.

[0027] Distributions C3 and / or C4 are preferably Gaussian distributions. For example, distribution C3 may be a Gaussian distribution with a mean of 400,000 to 2,500,000 and a standard deviation of 150,000 to 800,000 (e.g., segment size components 1-3, ss1-ss3, etc. in Table 1). Distribution C4 may be a Gaussian distribution with a mean of 15,000,000 to 60,000,000 and a standard deviation of 5,000,000 to 21,000,000 (e.g., segment size components 7-9, ss7-ss9, etc. in Table 1). In embodiments, a default threshold for the sum of the posterior probabilities of the segment sizes is 0.029 to 0.1527743, 0.029 to 0.033, 0.032 to 0.033, or about 0.1527743. In one embodiment, C3 is a Gaussian distribution with a mean of 400,000 to 800,000 and a standard deviation of 150,000 to 800,000 (e.g., segment size component 1, ss1, etc. in Table 1), and the default threshold is 0.02 to 0.032, for example, 0.032. In another embodiment, C3 is a Gaussian distribution with a mean of 800,000 to 1,800,000 and a standard deviation of 150,000 to 800,000 (e.g., segment size component 2, ss2, etc. in Table 1), and the default threshold is 0.032 to 0.033, for example, 0.033. In the embodiment, C3 is a Gaussian distribution having a mean of 1,800,000 to 2,500,000 and a standard deviation of 150,000 to 800,000 (e.g., segment size component 3, ss3, etc. in Table 1), and the default threshold is 0.03 to 0.033, for example, 0.033.

[0028] In one embodiment, C4 is a Gaussian distribution with a mean of 15,000,000 to 25,000,000 and a standard deviation of 5,000,000 to 21,000,000 (e.g., segment size component 7, ss7, etc. in Table 1), and the default threshold is 0.014 to 0.023, e.g., 0.014 or 0.023. In another embodiment, C4 is a Gaussian distribution with a mean of 25,000,000 to 55,000,000 and a standard deviation of 5,000,000 to 21,000,000 (e.g., segment size component 8, ss8, etc. in Table 1), and the default threshold is 0.029 to 0.031, e.g., 0.029 or 0.031.

[0029] In one embodiment, the default threshold is determined by comparing the sum of posterior probabilities for each distribution between resistant and susceptible samples in the training group and identifying a threshold that results in the highest proportion of samples being correctly identified as resistant and susceptible (i.e., a threshold that maximizes the accuracy of classification using the threshold). In another embodiment, the default threshold is determined by comparing the sum of posterior probabilities for each distribution between resistant and susceptible samples in the training group and identifying a threshold that results in the highest proportion of resistant samples being correctly identified as resistant, while no susceptible samples are misidentified as resistant (i.e., a threshold that maximizes the specificity of classification using the threshold—true negative rate—while having a 100% susceptibility—true positive rate).

[0030] In some embodiments, the method includes the steps of obtaining an aggregation metric that captures the contributions of multiple copy number features, and comparing the aggregation metric to a predetermined threshold, where an aggregation metric exceeding the predetermined threshold indicates the presence of lesion amplification. In some such embodiments, the aggregation metric is a copy number signature i(SbC i Exposure to (E i ) and E i The formula is as follows: PbC≒E×SbC (Formula 1) It is a value that satisfies the condition, and in the formula, E is a vector of size n containing coefficients E1, ..., n, and Ei is the exposure to signature i. PbC is a vector of size c ≥ 1, preferably 1 ≤ c ≤ 36, where each element in the vector, representing the sum of the posterior probabilities of each copy number event in the copy number profile, belongs to component C, and each component C is a distribution of values ​​for the copy number features. SbC is a matrix of size c × n, where each value represents the weight of component C in signature i.

[0031] SbC i The columns in the matrix SbC, denoted as such, contain the weights of all components C within the signature i. In some such embodiments, the default thresholds are E>0, E>0.005, E>0.01, E>0.015, or E>0.02. Preferably, the default thresholds are between 0 and 1%. In embodiments, the default threshold for E is obtained by comparing the values ​​of E for tolerable and susceptible samples within the training group.

[0032] The training group may include samples acquired from or derived from patients (e.g., organoids or spheroids). Drug-related susceptibility / resistance status may be known from (e.g., clinical trial data) or can be inferred by measuring the susceptibility of cells in the sample to the drug. For example, this may be done by determining the IC50 of the drug for each sample in the training group (e.g., using cell viability as a read) and classifying the n most susceptible samples (lowest IC50) as susceptible, where n may be selected based on the expected proportion of susceptible samples in the training group. Preferably, formula (1) is,

number

[0033] In the embodiment, the components include at least a component C1 which is a distribution of copy number change point values ​​centered around values ​​of 2 to 32. C1 may also be a distribution of copy number change point values ​​centered around values ​​of 4 to 30, 5 to 30, or 6 to 30. The distribution is preferably a Gaussian distribution such as a Gaussian distribution with a mean of about 28.7 (e.g., 28.7 ± 10%) and a standard deviation of about 22.1 (e.g., 22.1 ± 10%) (e.g., copy number change point component 7, cp7, etc. in Macintyre et al.), a Gaussian distribution with a mean of about 7.3 (e.g., 7.3 ± 10%) and a standard deviation of about 3.4 (e.g., 3.4 ± 10%) (e.g., copy number change point component 6, cp6, etc. in Macintyre et al.), or a Gaussian distribution with a mean of about 3.0 (e.g., 3.0 ± 10%) and a standard deviation of about 1.0 (e.g., 1.0 ± 10%) (e.g., copy number change point component 5, cp5, etc. in Macintyre et al.). When a single component is used, any arbitrary non-zero value can be used for the component weight, as the coefficients of the SbC vector capture the relative contribution of the component. Therefore, in such embodiments, exposure E can be calculated in the simplest case as the sum of the posterior probabilities of each event in the copy number profile belonging to the distribution described above. In other words, a weight of 1 can preferably be used for C1.

[0034] In embodiments, the components include at least one component C2 which is a copy number distribution centered around values ​​between 5 and 34. C2 may be a copy number distribution centered around values ​​between 6 and 34, 7 and 34, 6 and 32, 7 and 32, or 8 and 32. The distribution is preferably a Gaussian distribution such as a Gaussian distribution with a mean of about 30.9 (e.g., ±10%) and a standard deviation of about 23.1 (e.g., ±10%) (e.g., copy number component 8, cn8 in Macintyre et al.) or a Gaussian distribution with a mean of about 8.4 (e.g., ±10%) and a standard deviation of 3.5 (e.g., copy number component 7, cn7 in Macintyre et al.), where the weight of component C2 relative to the weight of component C1 is 2.5 to 0.3 (or 0.5 to 1.5, e.g., about 0.7).

[0035] In an embodiment, the components include at least one component C3 which is a distribution of segment sizes centered around values ​​of 100,000 to 4,000,000 base pairs, and at least one component C4 which is a distribution of segment sizes centered around values ​​of 12,000,000 to 80,000,000 bp. C3 may also be a distribution of segment sizes centered around values ​​of 200,000~4,000,000 bp, 300,000~4,000,000 bp, 400,000~4,000,000 bp, 100,000~3,000,000 bp, 200,000~3,000,000 bp, 300,000~3,000,000 bp, 400,000~3,000,000 bp, 100,000~2,500,000 bp, 200,000~2,500,000 bp, 300,000~2,500,000 bp, or 400,000~2,500,000 bp. C4 may be a segment size distribution centered around a value of 15,000,000 to 60,000,000. Preferably, components C3 and / or C4 are Gaussian distributions. For example, component C3 may be a Gaussian distribution with a mean of 400,000 to 2,500,000 and a standard deviation of 150,000 to 800,000 (e.g., segment size components 1-3, ss1-ss3, etc. in Macintyre et al.). Component C4 may be a Gaussian distribution with a mean of 15,000,000 to 60,000,000 and a standard deviation of 5,000,000 to 21,000,000 (e.g., segment size components 7-9, ss7-ss9, etc. in Macintyre et al.). The weight of component C3 relative to the weight of component C1 may be 3.5 to 0.1 (preferably 3.5 to 0.5, or 0.5 to 2.5, e.g., about 1). The weight of component C4 relative to the weight of component C1 may be 1 to 0.1 (preferably 1 to 0.2, e.g., about 0.3). In an embodiment, the component consists of one or more components C1. In an embodiment, the component includes one or more components C1 and one or more components C2. In an embodiment, the component includes one or more components C1, one or more components C3, and one or more components C4.In one embodiment, the component consists of one or more components C1, one or more components C2, one or more components C3, and one or more components C4.

[0036] In an embodiment, the components include up to seven components that are distributions of copy number change point values, all of which may be Gaussian distributions. In an embodiment, the components include up to eight components that are distributions of copy numbers, all of which may be Gaussian distributions. In an embodiment, the components include up to ten components that are distributions of segment sizes, all of which may be Gaussian distributions. In an embodiment, the components further include one or more components that are distributions of the number of breakpoints per xMB (x is, for example, 10), each of which may be a Poisson distribution. In an embodiment, the components include up to three components that are distributions of the number of breakpoints per xMB, all of which may be Poisson distributions. In an embodiment, the components further include one or more components that are distributions of the length of segments having a vibrational copy number, each of which may be a Poisson distribution. In an embodiment, the components include up to three components that are distributions of the length of segments having a vibrational copy number, all of which may be Poisson distributions. In an embodiment, the components further include one or more components that are distributions of the number of breakpoints per chromosome arm, each of which may be a Poisson distribution. In embodiments, the components include up to five components which are distributions of the number of breakpoints per chromosome arm, all of which may be Poisson distributions. In embodiments, the components include components 1 to 36 in Table 1, or distributions having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1. The corresponding elements of SbC for each component may be the corresponding weights specified in Table 1, or weights within 10%, 5%, 2%, or 1% of the weights in Table 1. The components may include components 1 to 36 in Table 1, and the corresponding elements of SbC may be the corresponding weights specified in Table 1. The components may consist of components 1 to 36 in Table 1, and the corresponding elements of SbC may be the corresponding weights specified in Table 1.The components may include components 1 to 36 in Table 1, or distributions having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1, and the corresponding elements of SbCi for each component may be the corresponding weights specified in Table 1, or weights within 10%, 5%, 2%, or 1% of the weights in Table 1, and SbC further includes weights for additional signatures specified in Table 2, weights within 10%, 5%, 2%, or 1% of the weights in Table 2.

[0037] Exposure to signature i may be advantageously calculated using all seven signatures in Tables 1 and 2, or the corresponding signature, with exposure Ei indicating the presence of lesion amplification being that of Table 1 (signature 6). In some such embodiments, the patient is an ovarian cancer patient, and the drug is an anthracycline, preferably doxorubicin. The signatures corresponding to those in Tables 1 and 2 may be signatures obtained by applying the method described in Macintyre et al. (2018) to different sets of copy number profiles and identifying the signatures obtained through this process that have the closest pattern of component contribution to the signatures in Macintyre et al. (2018) (Tables 1 and 2 below).

[0038] In a second embodiment, the present invention provides a method for predicting the response of cancer patients, such as ovarian cancer patients, to treatment with a DNA insertion agent, preferably an anthracycline such as doxorubicin, and this method is (a) A step of obtaining a tumor copy number profile for the patient, (b) Copy number profile, For each copy count event within the copy count profile, quantify multiple copy count features. To obtain an aggregated metric that captures the contributions of multiple copy number features, Comparing the aggregation scale to a default threshold. The step involves analyzing the patient, where if the aggregate scale exceeds a predetermined threshold, the patient is predicted to be resistant to treatment, and the aggregate scale is the copy number signature i(SbC i Exposure to (E i ) and E i The formula is as follows: PbC≒E×SbC (Formula 1) It is a value that satisfies the condition, and in the formula, The elements within PbC are the sum of the posterior probabilities of each copy number event in the copy number profile belonging to component C, where each component C is a distribution of values ​​for copy number features as defined in Table 1. The elements within SbC are the weights of each component C in the signature as defined in Tables 1 and 2, and E i This is an exposure to a signature with the weights specified in Table 1 (Signature 6). Includes steps.

[0039] In a third embodiment, the present invention provides a method for predicting the response of cancer patients to treatment, the method being To provide the whole genome sequence of the patient's tumor, the procedure involves sequencing the DNA obtained from the patient, The steps include obtaining a patient-specific tumor copy number profile from the whole genome sequence of the tumor, The step of using the tumor copy number profile thus obtained to carry out a first or second aspect of the present invention.

[0040] In some embodiments, sequencing includes extracting and / or amplifying DNA from a sample. The sample may be a tumor tissue sample containing circulating tumor DNA (ctDNA) or a liquid biopsy sample, such as a blood or plasma sample.

[0041] In some cases, the sequencing step may include next-generation sequencing (NGS), including Illumina® sequencing or Sanger sequencing. NGS provides the speed and accuracy required to detect mutants, for example, through whole-genome sequencing (WGS). Examples of NGS techniques include methods using synthetic sequencing, hybridization sequencing, ligation sequencing, pyrosequencing, nanopore sequencing, or electrochemical sequencing. In some embodiments, the sequencing is whole-genome sequencing. Advantageously, the sequencing may be shallow whole-genome sequencing. For example, shallow whole-genome sequencing with a depth of at least 2.5 million reads, preferably at least 2.7 million reads, may be used. In embodiments, the sequencing depth to be used is determined using the method described by Macintyre et al. (Trends Genet. 2016 Sep; 32(9): 530-542. doi: 10.1016 / j.tig. 2016.07.002.), which is incorporated herein by reference. The inventors have found that shallow whole-genome sequencing is sufficient to detect copy number variations with sufficient power to carry out the method of the present invention, especially when the sequencing depth is adapted to the expected ploidy and purity of the sample.

[0042] In some cases, the method of this aspect of the present invention further includes a step of preparing a DNA library from a sample (e.g., tumor sample, blood or plasma sample) obtained from a patient or from two or more patients prior to sequencing. Optionally, the library may be barcoded. In some cases, the method of this aspect of the present invention further includes a step of obtaining a sample from a patient prior to sequencing. For example, a blood sample or tissue biopsy may be collected from a patient diagnosed with or likely to have cancer. Additionally or alternatively, a sample containing circulating tumor DNA (ctDNA) (so-called liquid biopsy) may be obtained and used to provide the whole genome of a tumor for somatic mutant calling. Liquid biopsy techniques may be preferred when tissue biopsy is considered too invasive, or when the location of the tumor is unknown, or when the tumor is only suspected or in the context of tumor surveillance, including after surgical resection. The sample may be subjected to one or more extraction or purification steps, such as centrifugation, to obtain a substantially cell-free DNA source (e.g., a plasma sample). The samples are explicitly intended to be transported and / or stored (optionally, after freezing). Sample collection may occur at a location separate from the sequencing site, and / or computer-implemented method steps may occur at a location separate from the sample collection site and / or the sequencing site (for example, computer-implemented method steps may be performed using networked computers, such as with a “cloud” provider). Nevertheless, the entire method may, in some cases, be performed at a single location, which may be advantageous for “in-situ” classification or monitoring of cancer.

[0043] According to a fourth aspect, a system for predicting the treatment response of cancer patients is provided, and this system is At least one processor, The system comprises at least one non-temporary computer-readable medium containing an instruction, and when the instruction is executed by at least one processor, the system provides to at least one processor, (a) Obtain a tumor copy number profile for the patient, (b) Analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, wherein the at least one copy number feature is selected from copy number change points, segment size, and segment copy number, (c) If the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, the system causes the system to perform an action that predicts the patient is likely to be resistant to treatment with drugs that induce micronucleus formation. In an embodiment, when the instruction is executed by at least one processor, the system causes at least one processor to receive the whole genome sequence of the patient's tumor and to calculate a copy number profile from the whole genome sequence of the tumor.

[0044] In some embodiments, the system is intended for use in methods according to the first or second aspects of the present invention.

[0045] In a fifth embodiment, the present invention provides a non-temporary computer-readable medium for predicting the treatment response of cancer patients, the non-temporary computer-readable medium includes instructions, which, when executed by at least one processor, (a) Obtain a tumor copy number profile for the patient, (b) Analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, wherein the at least one copy number feature is selected from copy number change points, segment size, and segment copy number, (c) If the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, the system will perform actions including predicting that the patient is likely to be resistant to treatment with drugs that induce micronucleus formation. In some embodiments, the medium is for use in methods of the first or second aspects of the present invention.

[0046] In a sixth aspect, the present invention provides a method for predicting that a patient's tumor is resistant to anthracyclines such as doxorubicin, the method comprising the step of carrying out the method of a first aspect of the present invention, wherein the tumor is predicted to be resistant if the properties of at least one copy number feature indicate the presence of lesion amplification in the tumor genome. Preferably, the properties of at least one copy number feature are exposure (E) to a copy number signature as described herein, and the property is considered to indicate the presence of lesion amplification in the tumor genome if the exposure E is above a threshold. For example, the thresholds are E>0, E>0.01, E>0.02, or E>0.023. As described in detail herein, exposure to a copy number signature as described herein provides a reliable indicator that the tumor is resistant to anthracyclines, and in particular doxorubicin.

[0047] In a seventh aspect, the present invention provides a method for predicting whether a patient with a tumor would benefit from therapy with an anthracycline, preferably a DNA insertion agent such as doxorubicin, the method comprising the step of carrying out a method of a first or second aspect of the present invention, wherein the patient would not benefit from the therapy if the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome.

[0048] According to embodiments of any aspect of the present invention, if the characteristics of at least one copy number feature do not indicate the presence of lesion amplification in the tumor genome, the patient / tumor is predicted to be sensitive to the therapy. By limiting treatment with DNA intercarration therapy (e.g., anthracycline, preferably doxorubicin therapy) to cancer patients predicted to be sensitive to such therapy, the therapy is concentrated on those most likely to benefit from it, while excluding patients less likely to benefit from such therapy, for example, presumed non-responders from undesirable side effects associated with DNA intercarration therapy.

[0049] When the patient / tumor is predicted to be sensitive to the therapy, the method of the present invention Selecting patients for treatment with that therapy, and / or The present invention may further include administering a therapeutically effective dose of therapy to a patient in order to treat a tumor. When the patient / tumor is predicted to be resistant to the therapy, the present invention Selecting patients for treatment with an alternative therapy, and / or This may further include administering a therapeutically effective dose of another therapy to the patient to treat the tumor.

[0050] Other therapies may include, for example, paclitaxel. Paclitaxel is commonly administered to ovarian cancer patients in the UK.

[0051] Embodiments of the present invention will be described herein by reference to the accompanying drawings, but will not be limited thereto. However, various further aspects and embodiments of the present invention will be obvious to those skilled in the art in light of this disclosure.

[0052] The present invention includes, but is not limited to, combinations of the described embodiments and preferred features, where such combinations are clearly unacceptable or explicitly stated to be avoided. These and further embodiments and models of the present invention are described below in more detail, and with reference to the accompanying examples and figures. [Brief explanation of the drawing]

[0053] [Figure 1] This figure shows the doxorubicin response in primary ovarian cancer organoids, with the IC50 (drug concentration that induces a 50% reduction in cell viability, x-axis, scale = 10 μM) of 10 primary ovarian cancer organoids treated with doxorubicin. Each point represents the mean of three technical reproductions, and the two organoids with the lowest IC50 were considered sensitive based on the predicted sensitivity rate to doxorubicin in the patient population from which the organoids were derived. [Figure 2] This figure shows that the doxorubicin response correlates with copy number signature 6 exposure in patient-derived spheroids, where copy number signature (signature 6) exposure was calculated for 12 patient-derived spheroids from ascites (y-axis), and doxorubicin IC50 (drug concentration that induces a 50% reduction in cell viability) was measured for each of these samples, with the three samples having the lowest IC50 (the first three samples on the left, x-axis) representing the doxorubicin response in the patient population from which the spheroids were derived. Based on the predicted susceptibility rate, samples were considered susceptible. At the exposure > 0 cutoff, five samples were predicted to be susceptible to doxorubicin (54356, 80630, 80720, 54327, 119025 – the last two had higher IC50s than the three samples considered susceptible), and seven samples were predicted to be resistant (119016, 54075, 80601, 54289, 118902, 119178, 119120 – all of these had IC50s that indicated resistance), as shown in the figure. [Figure 3]This figure shows that copy number signature exposure predicts resistance to doxorubicin in a group of ovarian cancer patients, where copy number signature 6 exposure (y-axis) was calculated for 24 ovarian cancer patients in the CTCR-OV04 group, an ovarian clinical trial established to identify biological markers of response for doxorubicin. Patients were classified into four response categories: complete response, partial response (0 / 24 patients in this setting), stable, and progressive disease (x-axis). Using a cutoff for copy number signature exposure >0, progressive and stable cases were identified as resistant to doxorubicin with a sensitivity of 0.5. [Figure 4] The figure shows that copy number signature exposure predicts survival, with Kaplan-Meier curves generated separately for patients predicted to be resistant (signature 6 exposure > 0) and patients predicted to be susceptible. Patients predicted to be resistant were approximately five times more likely to experience disease progression compared to patients predicted to be susceptible, according to a Cox proportional hazards model (HR 5.2, p-value 0.0011, log-rank test) fitted with signature 6 exposure > 0 as a covariate. [Figure 5]Figure 1 shows sample details and workflow for analysis in the examples, where primary ovarian cancer organoids were derived from 11 patients, which were (a) sequenced to calculate copy number signature exposure and (b) treated with doxorubicin to determine the IC50 of each organoid, and the expected resistance rate to doxorubicin in the organoid population was used to define a cutoff for exposure to copy number signatures showing lesion amplification as described herein, the cutoff was validated using spheroids derived from ascites samples from 19 patients, here again spheroids The study found that (a) samples were sequenced to calculate copy number signature exposure and (b) treated in vitro with doxorubicin, and that exposure to signatures showing lesion amplification was associated with resistance to doxorubicin (Figure 2). This finding was validated using tissue-derived data from an ovarian cancer clinical study, in which samples from 53 patients were sequenced, of which 24 were selected after quality control for the calculation of copy number signature exposure, and this analysis showed that exposure to copy number signatures showing lesion amplification as described herein can be used to identify cases in which resistance to doxorubicin is present (Figures 3-4). [Figure 6A] Figure 6A shows that copy number signature exposure can be obtained from tumor tissue samples or ctDNA samples, with matched tumor tissue and blood sample ctDNA copy number profiles for patient 139 showing a high level of correspondence. [Figure 6B] Figure 6B shows the exposures for signatures 1-7 calculated from matched tumor tissue samples (left) and ctDNA samples (right) for patient 139, indicating high levels of response and, in particular, exposure to signature 6 > 0 in both cases (resulting in the patient being classified as resistant using either type of sample). [Modes for carrying out the invention]

[0054] In describing the present invention, the following terms are used and are intended to be defined as shown below.

[0055] When used herein, “and / or” shall be considered a specific disclosure of each of the two designated features or components, with or without the other. For example, “A and / or B” shall be considered a specific disclosure of (i) A, (ii) B, and (iii) A and B, as if each were presented individually herein.

[0056] When used herein, “computer implementation method” is deemed to mean a method involving the use of a computer, computer network, or other programmable device, and one or more features of the Method are implemented, in whole or in part, using a computer program.

[0057] When used herein in accordance with any aspect of the present invention, “patient” is intended to be equivalent to “subject,” and specifically includes both healthy individuals and individuals having a disease or disorder (e.g., a proliferative disorder such as cancer). Patients may be humans, pets (e.g., dogs or cats), laboratory animals (e.g., mice, rats, rabbits, pigs, or non-human primates), animals with xenografted or xenotransplanted tumors or tumor tissue (e.g., from human tumors), or livestock (e.g., pigs, cattle, horses, or sheep). Preferably, patient is a human patient. In some cases, patient is a human patient who has been diagnosed with cancer, is suspected of having cancer, or is classified as being at risk of cancer progression.

[0058] As used herein, “Sample” may be a biological sample, such as a cell-free DNA sample, a cell (including circulating tumor cells) or tissue sample (e.g., biopsy), a biological fluid, or an extract (e.g., a DNA extract obtained from a subject). In particular, a sample may be a tumor sample, a biological fluid sample containing DNA or cells, a blood sample (including plasma or serum samples), a urine sample, a cervical smear, or an ascites sample. Urine, ascites, and cervical smears are known to contain cells and therefore may provide suitable samples for use according to the present invention. Other sample types suitable for use according to the present invention include microneedle aspiration, lymph node samples (e.g., aspiration or biopsy), resection margins, bone marrow, or other tissue from the tumor microenvironment in which traces of tumor DNA may or may be found. A sample may be freshly obtained from a subject (e.g., blood collection) or may have been processed and / or stored before the determination is made (e.g., subjected to one or more purification, concentration, or extraction steps, including freezing, fixing, or centrifugation). For example, the sample may be a formalin-fixed tumor sample. The sample may be derived from one or more of the above biological samples through a concentration or amplification process. For example, the sample may include a DNA library produced from a biological sample, which may optionally be a barcoded or separately tagged DNA library. Multiple samples may be collected from a single patient, for example, sequentially during a series of treatments. Furthermore, multiple samples may be collected from multiple patients. In embodiments, the sample is a sample containing tumor cells. Preferably, such a sample has a tumor purity of at least 30%, at least 35%, at least 40%, at least 45%, or at least 50% (tumor purity may be quantified as the percentage of cells in the sample that are tumor cells). Advantageously, the sample has a tumor purity of at least 40%. Although we do not wish to be constrained by theory, copy number profiles produced from samples with lower tumor purity may be less suitable for the purposes of the present invention because signals corresponding to the tumor genome may be lost among signals from the genomes of other cells.

[0059] A “copy number profile” refers to the quantification of the number of copies for each of several segments of a genome sequence. In the context of this disclosure, a copy number profile is preferably a genome-wide copy number profile. A copy number profile is typically obtained by sequencing a sample of genomic DNA (or, as described above, a DNA library derived therefrom, including a sample of DNA derived from genomic DNA by fragmentation, such as cell-free DNA) and quantifying the number of copies per segment of the genome sequence (e.g., per bin, where a bin may be, for example, a 30kb region), as is known in the art. A copy number event refers to a segment of the genome, which has a copy number associated with it, and the copy number associated with a segment is different from the copy number associated with the immediately adjacent segment. The reason why the copy number associated with a segment differs from the copy number associated with the immediately adjacent segment is that the segments surrounding the segment are associated with different copy numbers, the segments surrounding the segment are not associated with any copy number (e.g., data for the segment is missing or of insufficient quality), or a combination of both (e.g., a segment may be surrounded by segments that are associated with different copy numbers on one hand, and segments that are not associated with any copy number on the other). In other words, a copy number event refers to the longest continuous segment of a copy number profile to which a single copy number is associated, each. For this reason, copy number events are also referred to as “segments” in this specification. In embodiments, a copy number profile includes up to 350 segments (copy number events), up to 300 segments, or up to 250 segments. Preferably, the copy number profile includes up to 250 segments (copy number events). These numbers may be particularly useful when looking at copy number profiles on a human genome scale. Although we do not wish to be constrained by theory, a larger number of segments is thought to indicate undesirable DNA degradation (e.g., formalin-mediated DNA degradation).

[0060] A “tumor copy number profile” refers to a copy number profile associated with a tumor genome. Tumor copy number profiles can be obtained by sequencing a sample of genomic DNA that is primarily derived from or presumed to be primarily derived from tumor cells (or, as described above, a DNA library derived therefrom, or a sample of DNA derived from genomic DNA by fragmentation). Those skilled in the art will understand that such a sample may be contaminated with non-tumor DNA, and that this contamination can be minimized through processing of the sample or processing of the sequencing data, as is known in the art. Preferably, a tumor copy number profile is a copy number profile obtained by sequencing a sample of genomic DNA derived from tumor cells. In other words, a tumor copy number profile is preferably obtained using a sample containing tumor cells.

[0061] The inventors have discovered that the presence of genome-wide characteristics of copy number features associated with lesion amplification in cancer is an indicator of a high likelihood of resistance to chemotherapy that induces micronucleus formation.

[0062] Copy count characteristics Copy number (CN) features are the properties of copy number events observable within a copy number profile. Copy number features may include segment size (length of each genomic segment within the copy number profile), breakpoints per xMB (number of genomic breaks appearing in a slide window across the copy number profile, the window is preferably 10MB, and the copy number profile is preferably genome-wide), change point copy number (absolute difference in copy number between adjacent segments within the copy number profile), segment copy number (observed absolute copy number state of each segment, also referred to herein as “copy number” or “absolute copy number”), breakpoints per chromosome arm (number of breaks occurring per chromosome arm), and the length of a segment having an oscillating copy number (number of consecutive segments alternating between two copy number states, rounded to the nearest integer copy number state).

[0063] CN features may be observed on a genome-wide basis for a single sample or a group of samples. In the context of this disclosure, the term “genome-wide” refers to the assessment of features across copy number profiles representing a substantial portion of the genome. For example, a substantial portion of the genome may include, or be derived from, a single chromosome, multiple chromosomes, or a portion of the genome, as determined by parameters of the sequencing process used, such as sequencing depth. In practice, those skilled in the art will understand that even whole-genome sequencing protocols may not be able to accurately capture all sequences in the genome, especially at low sequencing depths.

[0064] Copy number features that can be characterized on a genome-wide basis and may be used as indicators of lesion amplification include one or more of the following: segment copy number (e.g., high segment copy numbers, such as copy numbers greater than 5, may indicate the presence of lesion amplification), copy number change points (e.g., the presence of high copy number change points, such as copy number change points greater than 2, preferably greater than 4, may be associated with the presence of lesion amplification), and segment size (e.g., a combination of the presence of small and large segments, such as segment sizes less than 4,000,000 bp and segment sizes greater than 12,000,000 bp, may indicate the presence of lesion amplification). A particularly advantageous feature that may be used as an indicator of lesion amplification is the presence of high copy number change points. The genomic characterization of copy number features can be assessed by quantifying the copy number features for each copy number event in the copy number profile and obtaining one or more aggregated measures for the copy number profile. One or more aggregated measures may be obtained for each copy number feature. Alternatively, one or more aggregated measures that capture the contributions of multiple features may be obtained. One such measure is exposure to a signature representing the genomic imprint of an obvious putative mutation process, e.g., exposure to the signature described by Macintyre et al. (Nat Genet. 2018 Sept; 50(9): 1262-1270), which is incorporated herein by reference. In particular, exposure to signature 6, as disclosed by Macintyre et al. (2018) and further described below, can be calculated as a single measure indicating the presence of lesion amplification.

[0065] As described by Alexandrov et al. (Cell Rep. January 31, 2013; 3(1):246-59. doi:10.1016 / j.celrep.2012.12.008.), exposure to a mutation signature represents the number of mutants in a particular genome that result from that signature. The mutation process signature is the probability of the mutation process that causes each of the possible mutant types in the mutant catalog. In Macintyre et al. (2018), mutant types in the mutant catalog were defined as individual components of a mixed model fitted to a distribution of values ​​obtained for each of six copy number features across a specified genome-wide copy number profile from the HGSOC population. The copy number signatures (i.e., the copy number change process signatures) are obtained using these, and they capture the probability of the copy number change process that causes copy number events distributed according to each of the copy number feature components. In this context, exposure captured the strength of evidence for the presence of copy number change events attributable to the signature. Signature 6, as described by Macintyre et al. (2018), is thought to be associated with a mutation process that results in lesion amplification, which manifests as a significant contribution of copy number feature components that capture both high copy number change points, high copy numbers, and both small and large segment sizes. Lesion amplification is a copy number anomaly of a limited size. For example, lesion amplification may be associated with a region ≤ 3 Mb.

[0066] Copy number signature 6 was obtained as described in Macintyre et al., Nat Genet. 2018 Sept;50(9):1262-1270. Briefly, six copy number features (segment size, number of breakpoints per 10 Mb, change point copy number, segment copy number, number of breakpoints per chromosome arm, and segment length with vibrational copy number, as described above) were calculated for 91 samples within the BriTROC-1 HGSOC group. Mixture modeling was applied (using the FlexMix V2 package in R) to separate the copy number feature distributions from all samples into Poisson or Gaussian mixtures. Gaussian mixtures were used for the distributions representing segment size, change point copy number, and segment copy number. Poisson mixtures were used for the distributions representing the number of breakpoints per 10 MB, segment length with vibrational copy number, and number of breakpoints per chromosome arm. This resulted in a total of 36 mixed components.

[0067] For each copy number event, the posterior probability belonging to one component was calculated. For each sample, these posterior event vectors were summed to obtain a sum vector of posterior probabilities. All posterior sum vectors were combined in a patient × component posterior probability sum matrix. To identify copy number signatures, this matrix was subjected to non-negative matrix factorization (NMF) using the NMF package in R, according to the Brunet algorithm specification. This allowed for the convolution of the patient × component posterior sum matrix into the patient × signature matrix and signature × component matrix, identifying seven CN signatures, as well as their definitive features (as indicated by the coefficients of the signature × component matrix) and exposure within each sample (coefficients of the patient × signature matrix).

[0068] The component weights identified by NMF indicated which patterns of global or local copy number changes dictated each signature. CN signature 6 exhibited extremely high copy number states and high copy number change points from small high copy number segments scattered within larger, lower copy segments. This suggests a mutation process that results in lesion amplification.

[0069] Exposure to copy number signatures can be calculated for novel samples (i.e., samples other than those used to derive the signature), as described by Macintyre et al. (2018). In particular, copy number signature exposure is, The calculation of the coefficients of the patient × component vector (PbC) for a particular sample, or the PbC matrix for each of a set of samples, wherein the coefficients of the PbC vector for a particular sample can be calculated as the sum of the posterior probabilities of each copy number event belonging to one component (i.e., for each component and each sample, this is the sum of the posterior probabilities of events belonging to that component over all events for the sample), and Perform matrix decomposition to identify the values ​​(or vectors of values) that satisfy the following equation: PbC ≈ PbS × SbC (or, in practice, PbC = PbS × SbC + ε, where ε is the remainder term to be minimized), where SbC is the row of the signature × component matrix corresponding to signature 6, and PbS is the patient × signature value (or vector of values). It can be calculated by [method].

[0070] For example, exposure to multiple copy number signatures, such as any of signatures 1-7 by Macintyre et al. (2018), or corresponding signatures identified by applying the process described by Macintyre et al. (2018) (as described above) to different sets of tumor samples (copy number profiles), can be calculated using the corresponding rows of the signature × component matrix. In embodiments, copy number signature exposure is calculated using the LCD function in the YAPSA package within Bioconductor (https: / / rdrr.io / bioc / YAPSA / f / README.md). As those skilled in the art will understand, applying the method described by Macintyre et al. (2018) to different sets of copy number profiles (e.g., from samples from different groups, some containing some or all of the copy number profiles used in Macintyre et al. (2018), and some not) can result in different values ​​of the signature × component matrix. However, while we do not wish to be bound by theory, the signatures of Macintyre et al. (2018) are thought to reflect the underlying biological processes acting within the tumor. As such, while specific values ​​of the signature × component matrix may differ, such processes are expected to identify signatures that can match the signatures of Macintyre et al. (2018) (including at least signature 6) with respect to their patterns of contribution from copy number feature components. Such agreements are likely to be particularly evident when the method is applied to copy number profiles from ovarian cancer samples. Therefore, signatures corresponding to the signatures in Macintyre et al. (2018) refer to signatures identified using the processes described above, using the set of copy number features as described herein, and preferably using the set of components as described herein, for a set of copy number profiles from ovarian cancer samples.

[0071] In embodiments, exposure (E) to a copy number signature as described herein, such as signature 6 or a simplified version of signature 6 as disclosed by Macintyre et al. (2018), is used as a single measure indicating the presence of lesion amplification in a particular copy number profile, and exposure to copy number signature i (from a set of n signatures, where n can be, for example, 7, and the signatures can include any or all of the signatures disclosed by Macintyre et al. (2018), or corresponding signatures) is given by the following equation: PbC ≈ E × SbC (Equation 1) for a value E that satisfies i wherein E is a vector of size n that includes the coefficient E 1,…,n and E i is the exposure to signature i, PbC is a vector of size c from 1 to 36, and each value representing the sum of the posterior probabilities of each copy number event within the copy number profile belongs to component C, and each component C is a distribution of values for a copy number characteristic, SbC is a matrix of size c × n, where c is from 1 to 36, and each value represents the weight of component C within copy number signature i as described herein.

[0072] Preferably, exposure (Ei) to signature i is calculated using all 7 signatures within Macintyre et al. (2018), and the exposure Ei indicating the presence of lesion amplification is that of signature 6 within Macintyre et al. (2018), or a corresponding signature. In particular, in an advantageous embodiment, n = 7, and each of the 7 signatures corresponds to a signature within Macintyre et al. (2018) (i.e., it is either identical or can be a subset of the components within Macintyre et al. (2018) and / or have slightly different parameters due to fitting the distribution of copy number characteristics to multiple sets of samples) and is obtained through the process described herein, and the corresponding signature can be identified by those similar patterns of component contributions).

[0073] For example, the corresponding signature is: (I) To quantify copy number features for a set of tumor copy number profiles (preferably the tumor copy number profiles are from ovarian tumor cells), wherein the features are the segment size, the number of breakpoints per x MB (preferably 10 MB), the change point copy number, the segment copy number, the number of breakpoints per chromosome arm, and the length of the segment having the vibrational copy number. (ii) Identifying a mixture of statistical distributions to fit the distribution of each of these features across all copy number profiles, preferably a Gaussian distribution for segment size, change point copy number, and segment copy number, and a Poisson distribution for the number of breakpoints per xMB, segment length having vibrational copy number, and breakpoints per chromosome arm, preferably the number of distributions fitted for each feature being 7 distributions for copy number change points, 8 distributions for copy number, 10 distributions for segment size, 3 distributions for breakpoints per xMB, 5 distributions for breakpoints per chromosome arm, and 3 distributions for vibrational length. (iii) For each copy number event in each copy number profile, calculate the posterior probability belonging to each of the distributions identified in step (ii). (iv) For each copy number profile, sum the values ​​in (iii) for each component (i.e., one value per component for each copy number profile), (v) To obtain a matrix of exposure vectors to a set of signatures, and weights for each distribution (also referred to herein as “components”) and signature, non-negative matrix factorization (NMF) can be applied to the matrix obtained as a result of step (iv). Signatures corresponding to those in Tables 1 and 2 can be identified as signatures obtained in step (v) that exhibit a similar pattern of component contributions by establishing agreement between components in Table 1 and components identified in step (ii) (not all components in Table 1 have corresponding components in step (ii), and vice versa), and by comparing the weights obtained in step (v) with the weights for the corresponding components in Tables 1 and 2 based on the components that are found to have corresponding components. For example, if the components identified in step (ii) include fewer than 36 components, and some or all of them may match one of the 36 components in Table 1, then only the weights corresponding to these matched components in Tables 1 and 2 may be used. Similarly, if the component identified in step (ii) contains more than 36 components, only the weights in Tables 1 and 2 for components that can be matched to the component identified in step (ii) may be used.

[0074] Preferably, formula (1) is,

number

[0075] In one embodiment, the copy number profile is exposure E iIf the threshold is exceeded, it is considered evidence of exposure to signature i. The threshold can be predetermined, for example, 0, 0.005 (0.5%), 0.01 (1%), etc. Preferably, the threshold is 0 to 1% (i.e., 0 to 0.01). In embodiments, the threshold can be determined by comparing exposure values ​​in samples from patients with different phenotypes. For example, the threshold can be determined by comparing exposure to signature i (E) in one or more patients who are susceptible to the drug. i ) to calculate exposure to signature i (E) for one or more patients who are resistant to the drug. i This can be determined by calculating the threshold level of exposure and identifying the level of exposure that appropriately classifies a patient as susceptible or resistant. Whether the classification is appropriate may depend on several parameters, such as the treatment options that come to mind. Favorably, the threshold level of exposure may be identified as the level of exposure below which all susceptible patients fall. This can reduce the risk of susceptible patients being classified as resistant and therefore not receiving treatments that may have been effective.

[0076] The components include at least component C1, which is a distribution of copy number change point values ​​centered around values ​​between 2 and 32 (preferably 4 to 30, 5 to 30, or 6 to 30). The distribution is preferably a Gaussian distribution such as a Gaussian distribution with a mean of about 28.7 (e.g., 28.7 ± 10%) and a standard deviation of about 22.1 (e.g., 22.1 ± 10%) (e.g., copy number change point component 7, cp7, etc. in Macintyre et al.), a Gaussian distribution with a mean of about 7.3 (e.g., 7.3 ± 10%) and a standard deviation of about 3.4 (e.g., 3.4 ± 10%) (e.g., copy number change point component 6, cp6, etc. in Macintyre et al.), or a Gaussian distribution with a mean of about 3.0 (e.g., 3.0 ± 10%) and a standard deviation of about 1.0 (e.g., 1.0 ± 10%) (e.g., copy number change point component 5, cp5, etc. in Macintyre et al.). When a single component is used, any arbitrary non-zero value can be used for the component weight, since the coefficients of the SbC vector capture the relative contributions of the components. Thus, in such embodiments, exposure E can be calculated in the simplest case as the sum of the posterior probabilities of each event in the copy number profile belonging to the distribution described above. In other words, a weight of 1 can preferably be used for C1.

[0077] In embodiments, the components include at least one component C2 which is a distribution of copy number change point values ​​centered around values ​​of 5 to 34 (preferably 6 to 34, 7 to 34, 6 to 32, 7 to 32, or 8 to 32). The distribution is preferably a Gaussian distribution such as a Gaussian distribution with a mean of about 30.9 (e.g., ±10%) and a standard deviation of about 23.1 (e.g., ±10%) (e.g., copy number component 8, cn8 in Macintyre et al.) or a Gaussian distribution with a mean of about 8.4 (e.g., ±10%) and a standard deviation of 3.5 (e.g., copy number component 7, cn7 in Macintyre et al.), where the weight of component C2 relative to the weight of component C1 is 2.5 to 0.3 (or 0.5 to 1.5, e.g., about 0.7).

[0078] Therefore, the components consist of 100,000 to 4,000,000 base pairs (preferably 200,000 to 4,000,000 bp, 300,000 to 4,000,000 bp, 400,000 to 4,000,000 bp, 100,000 to 3,000,000 bp, 200,000 to 3,000,000 bp, 300,000 to 3,000,000 bp, 400,000 to 3,000,000 bp, and 100,000 to 2,500,000 The system includes at least one component C3, which is a distribution of segment sizes centered around values ​​of bp (200,000~2,500,000bp, 300,000~2,500,000bp, or 400,000~2,500,000bp), and at least one component C4, which is a distribution of segment sizes centered around values ​​of 12,000,000~80,000,000bp (preferably 5,000,000~60,000,000). Preferably, components C3 and / or C4 are Gaussian distributions. For example, component C3 may be a Gaussian distribution with a mean of 400,000~2,500,000 and a standard deviation of 150,000~800,000 (e.g., segment size components 1-3, ss1-ss3, etc. in Macintyre et al.). Component C4 may be a Gaussian distribution having a mean of 15,000,000 to 60,000,000 and a standard deviation of 5,000,000 to 21,000,000 (e.g., segment size components 7-9, ss7-ss9, etc., within Macintyre). The weight of component C3 relative to the weight of component C1 may be 3.5 to 0.1 (preferably 3.5 to 0.5, or 0.5 to 2.5, e.g., about 1). The weight of component C4 relative to the weight of component C1 may be 1 to 0.1 (preferably 1 to 0.2, e.g., about 0.3).

[0079] In one embodiment, the component consists of one or more components C1. In another embodiment, the component includes one or more components C1 and one or more components C2. In another embodiment, the component includes one or more components C1, one or more components C3, and one or more components C4. In yet another embodiment, the component includes one or more components C1, one or more components C2, one or more components C3, and one or more components C4. In yet another embodiment, the component consists of one or more components C1, one or more components C2, one or more components C3, and one or more components C4.

[0080] In embodiments, the components include 1, 2, 3, 4, 5, 6, 7, or up to 7 components, which are distributions of copy number change point values, and all of these may be Gaussian distributions. In embodiments, the components include 1, 2, 3, 4, 5, 6, 7, 8, or up to 8 components, which are distributions of copy values, and all of these may be Gaussian distributions. In embodiments, the components include 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, or up to 10 components, which are distributions of segment sizes, and all of these may be Gaussian distributions. In embodiments, the components further include one or more components, which are distributions of the number of breakpoints per xMB (x is, for example, 10), and each such distribution may be a Poisson distribution. In embodiments, the components include up to 3 components, which are distributions of the number of breakpoints per xMB, and all of these may be Poisson distributions. In embodiments, the components further include one or more components, which are distributions of the length of segments having vibrational copy numbers, and each such distribution may be a Poisson distribution. In embodiments, the components include up to three components which are distributions of segment lengths having vibrational copy numbers, all of which may be Poisson distributions. In embodiments, the components further include one or more components which are distributions of breakpoint numbers per chromosome arm, each of which may be a Poisson distribution. In embodiments, the components include up to five components which are distributions of breakpoint numbers per chromosome arm, all of which may be Poisson distributions.

[0081] In some embodiments, the components include components 1 to 36 as defined in Table 1. In some embodiments, the components include components 1 to 36 as defined in Table 1, and the corresponding coefficients in the SbC matrix are the corresponding weights as defined in Table 1. In Table 1, N(m,sd) refers to a Gaussian distribution with mean = m and standard deviation = sd, and P(l) is a Poisson distribution at λ = l. In some embodiments, the components include components 1 to 36 in Table 1, or distributions having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1. In some such embodiments, the components have the corresponding weights as defined in Table 1, or weights within 10%, 5%, 2%, or 1% of the weights in Table 1.

[0082] Table 1. Components and weights for calculating exposure to copy number signatures in accordance with this disclosure. [Table 1-1] [Table 1-2] [Table 1-3]

[0083] In an embodiment, the components include components 1 to 36 as defined in Table 1, and exposure to a signature as described herein is calculated as E6 using all seven signatures in Table 2 (T1 refers to Table 1, and signature numbers such as "Sig.1" correspond to the numbers in Macintyre et al. (2018)). In such an embodiment, the components include components 1 to 36 as defined in Table 1, and the corresponding coefficients for E6 in the SbC matrix are the corresponding weights as defined in Table 1, and E in the SbC matrix 1-5、7The corresponding coefficients for are the corresponding weights specified in Table 2. In some such embodiments, the components include components 1 to 36 in Table 1, or distributions having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1. In some such embodiments, the components have the corresponding weights specified in Tables 1 and 2, or weights within 10%, 5%, 2%, or 1% of the weights in Tables 1 and 2.

[0084] Table 2. Weights for calculating exposure to copy number signatures in accordance with this disclosure. [Table 2-1] [Table 2-2]

[0085] Cancers with high chromosomal instability Genomic and chromosomal instability is characterized by the presence of mutants within the genome, including single-nucleotide variants, copy number variants (including aneuploidy; also referred to as “copy number changes”), and / or chromosomal rearrangements (also referred to as structural changes when they involve a large portion of the genome, such as >1kb). Chromosomal instability is characterized by the presence of copy number changes and chromosomal rearrangements (including structural variants) within the genome. Genomic instability typically refers to the presence of single-nucleotide variants within the genome. Most cancers are thought to exhibit some degree of chromosomal instability. However, some cancers are considered to have high chromosomal instability, exhibiting an unusually high frequency of copy number changes and chromosomal rearrangements within their genomes. In particular, within the context of this invention, cancers with high chromosomal instability may have a large number of copy number polymorphism events within their genomes. A copy number polymorphism event is the presence of genomic selection in several repeats that differ from the expected number in the wild-type genome (i.e., 2 in the case of diploid organisms such as humans). While we do not wish to be constrained by theory, the present invention is considered particularly useful in the context of cancers with high chromosomal instability, as their genomes are most likely to show genomic-scale indications of lesion amplification that may exhibit resistance to micronucleus formation. Cancers with high chromosomal instability include, but are not limited to, lung cancer, esophageal cancer, pancreatic cancer, breast cancer, and ovarian cancer, as well as all epithelial cancers (epithelial malignancies), sarcomas, and glioblastomas.

[0086] In embodiments, the sample according to the present invention is from a patient diagnosed with or likely to have an epithelial malignancy. In embodiments, the sample according to the present invention is from a patient diagnosed with or likely to have lung cancer, esophageal cancer, pancreatic cancer, sarcoma, glioblastoma, breast cancer, or ovarian cancer. Preferably, the patient has been diagnosed with or likely to have ovarian cancer such as HGSOC. In embodiments, the patient has been diagnosed with or likely to have breast cancer such as triple-negative breast cancer.

[0087] It is particularly advantageous that the method of the present invention is used in relation to patients who have not yet received therapy with one or more DNA insertion agents. Preferably, the patient has not yet been treated with anthracyclines such as doxorubicin. In embodiments, the subject has not previously received chemotherapy. In embodiments, the patient has preferably previously received chemotherapy with agents other than anthracyclines. For example, the patient may have received chemotherapy with carboplatin and / or paclitaxel. In embodiments, the patient has undergone tumor debulking surgery. In embodiments, the patient has relapsed or has not responded to previous chemotherapy. In embodiments, the patient has been diagnosed with cancer that is resistant to platinum-based chemotherapy. In some such embodiments, the patient has not received further chemotherapy after being diagnosed with platinum resistance. For example, a patient may be diagnosed with platinum resistance if their cancer relapses within a certain period (e.g., 6 months) after receiving platinum-based therapy. In particular, the patient may not have received paclitaxel (monotherapy) and / or doxorubicin (monotherapy) after being diagnosed with platinum resistance. In other such embodiments, the patient may have received chemotherapy with a drug other than anthracyclines, such as paclitaxel, after being diagnosed with platinum resistance. In other embodiments, the patient is diagnosed with cancer that is sensitive to platinum-based chemotherapy. For example, a patient may be diagnosed with platinum sensitivity if their cancer does not recur within a certain period (e.g., 6 months) after receiving platinum-based therapy. In some such embodiments, the patient may not have received platinum-based chemotherapy after being diagnosed with platinum sensitivity.

[0088] In embodiments, the patient may be an ovarian cancer patient with homologous recombination repair deficiency, for example, a patient with a somatic or non-somatic BRCA1 / 2 mutant. Alternatively, the patient may be an ovarian cancer patient without homologous recombination repair deficiency. The patient may not be identified as having a somatic or non-somatic BRCA1 / 2 mutant, and / or may be considered to have wild-type BRCA1 and BRCA2 genes. Ovarian cancer patients identified as having a BRCA1 / 2 mutant are considered to be slightly more likely to benefit from doxorubicin therapy. However, for patients whose BRCA1 / 2 status is unknown or who are considered to be wild-type, there is currently no reliable way to predict whether the patient will benefit from treatment with chemotherapy such as doxorubicin. Such patients are in a position to benefit the most from the present invention. However, even patients who are known to have a BRCA1 / 2 mutant may benefit from the present invention due to the weak predictive power of the presence of a BRCA1 / 2 mutant.

[0089] therapy Micronuclei are chromatin-containing structures that form when a chromosome or a cross-section of a chromosome is not incorporated into the nucleus of either daughter cell during cell division. Micronuclei are thought to be associated with genotoxic events, DNA damage, and chromosomal instability. It has been hypothesized that micronucleus formation, when it does not lead to cell death, constitutes part of the process of cell transformation (Hintzsche et al., Scientific Reports, (2018) 8:3371).

[0090] Several chemotherapy therapies are known to induce micronuclei. These include genotoxic therapies, and especially genotoxic chemotherapy, including therapies that act as DNA insertion agents (such as anthracyclines), therapies that act as spindle toxins (such as vinblastine), therapies that stall the replication fork (such as alkylating agents, e.g., methylmethanesulfonate (MMS)), and DNA crosslinking agents (such as mitomycin C (MMC)) (Hintzsche et al., Scientific Reports, (2018) 8:3371). DNA insertion agents are complexes that can interact with DNA by inserting between base pairs. Several DNA insertion agents, including doxorubicin, daunorubicin, and dactinomycin, are used as chemotherapy in the treatment of cancer. In particular, DNA insertion agents that act as topoisomerase II toxins can induce micronucleus formation. Topoisomerase II toxins are complexes that block the action of topoisomerase II, preventing it from performing its normal function during the cell cycle. These include anthracyclines such as doxorubicin, daunorubicin, epirubicin, and idarubicin. In fact, there is long-standing evidence that anthracyclines can induce micronucleus formation (see, e.g., Bhuyan et al., Cancer Research 43, 5293-5297, 1983). Spindle toxins are complexes that interfere with the function of the mitotic spindle. Spindle toxins include vinca alkaloids such as vinblastine. Alkylating agents are complexes that add alkyl groups to guanine bases of DNA molecules. These include alkyl sulfonates, nitrogen mustard, ethyleneimine, nitrothrea, and triazines. DNA crosslinkers are drugs that react with two DNA nucleotides to form a covalent bond between them. Examples of DNA crosslinkers include cisplatin, oxaliplatin, mitomycin C, and nitrogen mustard.

[0091] Doxorubicin is widely used as an anti-cancer chemotherapy drug, including for the treatment of breast cancer, bladder cancer, sarcoma, lymphoma, and leukemia. Doxorubicin is available in liposomal (i.e., liposomal-encapsulated) and non-liposomal forms. Liposomal doxorubicin is available in a pegylated liposomal-encapsulated form, known in the market as Doxil. Doxorubicin is also available in a non-pegylated liposomal form, known in the market as Myocet. In the context of this disclosure, references to “doxorubicin” encompass all forms of doxorubicin. In particular, the term “doxorubicin” expressly encompasses the liposomal (pegylated and non-pegylated) and non-liposomal forms of doxorubicin unless otherwise specified.

[0092] The present invention provides a method for predicting whether a cancer patient is likely to respond to micronucleus-inducing chemotherapy, and a method for treating a patient who has cancer or has been diagnosed as likely to have cancer, wherein a series of treatments may be adapted depending on whether the patient is predicted to be likely to respond to micronucleus-inducing chemotherapy. As such, the step of predicting whether a cancer patient is likely to respond to chemotherapy may be advantageously performed in relation to (i) a patient who has not yet received any chemotherapy, (ii) a patient who has not yet received chemotherapy with a micronucleus-inducing agent, (iii) a patient who has not yet received chemotherapy with a DNA insertion agent, (iv) a patient who has not yet received chemotherapy with anthracyclines, and / or (v) a patient who has not yet received chemotherapy with doxorubicin. A method of treating a patient may include treating the patient with a micronucleus-inducing agent, preferably a DNA insertion agent, advantageously an anthracycline such as doxorubicin, if the patient is predicted to be unlikely to be resistant to these agents. Methods of treating a patient may include treating the patient with a non-micronucleus-inducing agent, preferably one that is not a DNA insertion agent (e.g., not an anthracycline such as doxorubicin), if the patient is predicted to be highly likely to be resistant to a micronucleus-inducing agent. Methods of treating a patient may include treating the patient with a micronucleus-inducing agent, preferably one that is not a DNA insertion agent, advantageously an anthracycline such as doxorubicin, either alone or in combination with further agents. Further agents may be micronucleus-inducing agents or non-micronucleus-inducing agents.

[0093] Furthermore, the step of predicting whether a cancer patient is likely to respond to chemotherapy may be repeated throughout the patient's entire treatment regimen. For example, this step may be performed before initiating chemotherapy with non-anthracycline drugs and after a patient's relapse following chemotherapy.

[0094] In some embodiments, the patient is an ovarian cancer patient, and the predictive step is performed before the patient receives chemotherapy, e.g., chemotherapy with carboplatin and paclitaxel, and / or tumor debulking surgery. In some embodiments, the patient is an ovarian cancer patient, and the predictive step is performed after the patient has relapsed following chemotherapy, e.g., chemotherapy with carboplatin and paclitaxel, and / or tumor debulking surgery. In some such embodiments, the patient has been diagnosed with platinum resistance, and the method of treating the patient includes treating the patient with a micronucleus-inducing agent (such as an anthracycline, preferably doxorubicin) if the patient is predicted to be unlikely to be resistant to this agent, or treating the patient with chemotherapy without this agent (e.g., paclitaxel monotherapy) if the patient is predicted to be likely to be resistant to this agent. In other embodiments, the patient has been diagnosed as platinum-sensitive, and the method of treating the patient includes treating the patient with a micronucleus-inducing agent (such as an anthracycline, preferably doxorubicin) in combination with or after a platinum-based therapy, if the patient is expected to be unlikely to be resistant to the agent, or treating the patient with chemotherapy that does not include the agent (e.g., cisplatin monotherapy) if the patient is expected to be likely to be resistant to the agent. In any embodiment described herein, the micronucleus-inducing agent (e.g., an anthracycline such as doxorubicin) may be administered as monotherapy (i.e., alone) or in combination with one or more other therapies (which may include other chemotherapy or other treatments such as radiotherapy).

[0095] "Predicting a cancer patient's response to a selected treatment" is intended to mean assessing the likelihood that a patient will experience a positive or negative outcome as a result of a particular treatment.

[0096] As used herein, “positive therapeutic outcome” means the increased likelihood that a patient will experience beneficial results from the chosen treatment, such as a reduction in tumor size, a “favorable” prognosis, or an improvement in disease-related symptoms and / or quality of life.

[0097] "Demonstrating a negative treatment outcome" is intended to mean an increased likelihood that the patient will not receive the aforementioned benefit of a positive treatment outcome.

[0098] The following are presented as examples only and should not be construed as limitations on the claims. [Examples]

[0099] Summary of Examples To test the hypothesis that genome-wide copy number features indicating lesion amplification are associated with resistance to drugs that induce micronucleus formation, primary ovarian cancer organoids were treated with doxorubicin, and the decrease in IC50 in cell viability was used as the response measure. Signature 6 exposure was calculated for each organoid, and a cutoff for exposure was defined using the expected resistance rate to doxorubicin in the organoid population. This cutoff was validated using patient-derived spheroids from ascites treated in vitro with doxorubicin, and it was found that exposure to the signature was associated with resistance to doxorubicin. Finally, retrospective validation in ovarian cancer clinical research showed that exposure to signature 6 can be used to identify cases resistant to doxorubicin using both tumor tissue and ctDNA samples. Sample details and the workflow for this analysis are illustrated in Figure 5.

[0100] material and method Ethical approval and clinical sample collection Patient clinical data and samples were collected as part of the promising Cambridge Translational Cancer Research Ovarian Study 04 (CTCROV04), approved by the Institutional Ethics Committee (REC08 / H0306 / 61). Patients provided written informed consent to participate in this study and to the use of their donated tissue for the clinical tests performed in this study.

[0101] Sample preparation for organoids Samples were obtained from surgical excision, ward drainage tubes, or surgical irrigation fluid. Solid tumors were evaluated by pathologists, and only tumor samples with ≥50% cellular density were attempted to grow.

[0102] Sample preparation for spheroids Ascites fluid was collected from patients in volumes of 100 ml to 2 L. This fluid was gently centrifuged at 800 x g for 5 minutes, and most of the supernatant was removed. The sample was filtered using an autoclaved muslin cloth, and the resulting fluid was filtered again using a 40 μm cell filter. The spheroids from the filter were then collected by washing with 10 ml of PBS and centrifuged at 1500 rpm for 5 minutes. The spheroid fragments were divided into two: a cell pellet for DNA extraction, and a resuspension of cells in filtered cell-free ascites supernatant and 8% DMSO for drug screening. For drug screening, the spheroids were thawed and left in culture medium overnight to ensure complete collection before dispensing.

[0103] Sample preparation for tissue FFPE tissue blocks were cut into 8 μm pieces, and tumor-rich areas were recovered by macro-dissection based on areas marked by the research pathologist in adjacent hematoxylin and eosin-stained regions. DNA was extracted from 3–10 pieces using the QIAmp DNA Microkit (Qiagen), with the following modifications to the original protocol: an additional incubation step with Buffer ATL at 95°C for 15 minutes was introduced before adding proteinase K; and paraffin was removed using the xylene / ethanol method.

[0104] Sample preparation for ctDNA Plasma samples were collected from patients with high serous ovarian dysplasia entrained in the OV04 prospective clinical study at Addenbrook Hospital in Cambridge, UK, which was approved by the local research ethics committee (REC reference number: 07 / Q0106 / 63, and NRES Committee East of England-Cambridge Central 03 / 018). Written informed consent was obtained from all patients, and blood samples were collected before and after the initiation of surgery or chemotherapy. The inventors focused on selected plasma time points collected before primary chemotherapy and before doxorubicin treatment (usually the second or third treatment). For a small number of patients, the plasma time point collected at the end of doxorubicin treatment was analyzed.

[0105] DNA was extracted from 2 or 4 mL of plasma using the QIAamp circulating nucleic acid kit (Qiagen) or QIAsymphony (Qiagen) according to the manufacturer's instructions. 10 μl of the extracted circulating nucleic acid was used as input for whole-genome library preparation using the ThruPLEX DNA-Seq (Takara) library preparation kit, with the following modifications: DNA shearing was not performed; 14 PCR cycles were applied; library purification using Ampure beads (Beckman Coulter) was performed separately for each sample; and elution was performed using 20 μl of Tris EDTA buffer. The generated libraries were quantified using the Fragment Analyser NGS kit (Agilent Technologies), diluted to 10 nmol / l and pooled in the same proportions. All libraries were sequenced using NovaSeq S2 (Illumina) in PE-150bp mode to achieve at least 80 mln reads per sample.

[0106] Organoid derivatives Tumor samples were washed in PBS, finely chopped into 2 mm pieces using a surgical scalpel, and incubated at 37°C for 1–2 hours with gentamicin (50 μg / ml), bovine serum albumin fraction V (1.5%), insulin (5 μg / mL), collagenase A (1 mg / mL), and hyaluronidase (100 U / mL). After incubation, the mixture was filtered, the cell suspension was allowed to settle, and washed with PBS. The ascites fluid was centrifuged at 450 g for 5 minutes. The cells were then washed with PBS and centrifuged at 400 g for 5 minutes. The isolated cells were resuspended in 7.5 mg / mL basement membrane matrix (Cultrex BME RGF type 2 (BME-2), Amsbio) plated as 20 μl droplets in a 6-well plate supplemented with complete medium. After polymerization of BME-2, complete medium was added, and the cells were kept at 37°C. Complete medium: AdDMEM / F12 medium supplemented with HEPES (1×, Invitrogen), Glutamax (1×, Invitrogen), Penicillin / Streptomycin (1×, Invitrogen), B27 (1×, Invitrogen), N2 (1×, Invitrogen), Wnt3a conditioning medium (25% v / v), RSPO1 conditioning medium (25% v / v), recombinant Noggin protein (100 ng / ml, Peprotech), epidermal growth factor (EGF, 10 ng / ml, Peprotech), fibroblast growth factor 10 (FGF10, 100 ng / ml, Peprotech), nicotinamide (1 mM, Sigma), SB431542 (0.5 μM, Cambridge Biosciences), and Y27632 (9 μM, Abmole).

[0107] Organoid culture The organoid culture medium was changed every other day. To subculture the organoids, the domes were scraped off and collected in Falcon tubes, TrypLE (Invitrogen) was added, and they were incubated at 37°C for approximately 10 minutes. The suspension was centrifuged at 800g for 2 minutes, and the cell pellet was resuspended in 7.5 mg / ml BME-2 plated as 20 μl droplets in a 6-well plate replenished with complete medium. After polymerization of the BME-2, complete medium was added, and the cells were incubated at 37°C. DNA was extracted from the cell pellet using the Qiagen Allprep DNA / RNA extraction kit according to the manufacturer's instructions.

[0108] In vitro doxorubicin treatment Using an Echo® 550 acoustic liquid handler instrument (Labcyte), an 8.5-log dilution series of doxorubicin, starting at 30 μM, was dispensed into 384-well plates and kept at -20°C until use. The organoid plates were allowed to settle, and 50 μl of suspension per well was added using a Multidrop® Combi Reagent Dispenser (Thermo-Fisher). After 5 days of drug incubation, cell viability was chemically analyzed using 30 μl of CellTiter-Glo® (Promega). Screening was performed using a technical triplicate. Response values ​​were normalized to be equivalent to the percentage of remaining viable cells using an untreated control group. To aid in dose-response curve fitting, dummy values ​​were added below and above the minimum and maximum dose ranges for each sample at 1e-03 μM for 100% viable cells and at 300 μM for 0% viable cells. Dose-response curves were fitted using a four-parameter log-logistic model, including the IC50 parameter used here. Fitting was performed in R using the drm function in the drc package (Ritz et al., 2015). For spheroid samples, IC50 values ​​were scaled by the reciprocal of tumor purity to account for increased cell viability due to normal cell contamination.

[0109] DNA sequencing Whole-genome sequencing libraries were prepared from 75 ng of DNA using the SMARTer Thruplex DNA-Seq (Takara) protocol. DNA from each sample was sheared using a Covaris LE220 (Covaris): duty cycle -30%, intensity -5.0, bursts per second -50, duration -120 seconds, peak injection power -180, temperature 20°C, water level -4. All samples underwent 7 PCR cycles. Library quality and quantity were assessed using a D5000 on a 4200 Tapestation according to supplier recommendations. The libraries were then pooled together and sequenced on a NovaSeq SP using PE-50 mode, aiming for 10 million reads per sample.

[0110] Quality control and sample filtering All samples in this study underwent a series of quality control (QC) tests, and unsuccessful samples were removed from further analysis. A summary of the filtering criteria and the number of samples retained for further analysis can be seen in Figure 5. The following QC criteria were used.

[0111] • Repeated samples (organoids, spheroids, and tissues): Each of the sample sets presented here contained at least one patient from whom multiple samples were profiled. In all cases, a single sample was selected for further analysis.

[0112] • Drug screening (organoid and spheroid samples): Samples showing greater than 20% standard deviation in cell viability across doses and concentrations greater than 3 were removed from downstream analysis.

[0113] • Copy number (tissue and ctDNA): Two main factors affect the ability to accurately quantify copy number signature exposure: 1) insufficient tumor genome sequence range, which leads to insufficient segmentation and signal loss; and 2) formalin-mediated DNA degradation, which leads to excessive segmentation of the genome and signal contamination. Factor 1 is generally caused by the low tumor purity of the sample. Therefore, we removed all tissue samples with a purity of less than 40%. To address Factor 2, we removed all samples (tissue and ctDNA) showing more than 250 segments across the genome.

[0114] Absolute copy number fitting Reads were aligned with the human genome assembly GRCh37 using BWA-MEM (Li & Durbin, 2009). Replication was marked using Picard (Broad Institute, 2018), and relative copy numbers were calculated using QDNAseq (Scheinin et al., 2014) with a 30kb bin size. For all bins across each sample, the absolute tumor copy number (the number of chromosomal copies of each DNA segment in tumor cells within the sample) was calculated. Each segmented relative copy number bin estimate j was converted from relative copy number (rCN) to absolute copy number (aCN):

number

[0115]

number

[0116]

number

[0117] Since organoid samples were assumed to be 100% pure, purity was fixed at 1, and the search was performed only across polyploid states. Purity / ploidy values ​​were excluded from consideration if they resulted in a fit indicating they were greater than 10 megabase pairs of a genome with homozygous loss. For tissue samples, an additional filter was used to remove fits that did not show at least one genomic segment across all integer copy number states from 1 to polyploidy.

[0118] Copy count signature calculation Copy number features were calculated from absolute copy number profiles, as detailed by Macintyre et al. (2018). Briefly, the absolute copy number profiles were summarized for each sample by calculating the genome-wide distribution of six features associated with copy number (CN) events: • Segment size - the length of each genome segment; • Number of breakpoints per 10MB - the number of genome breaks that appear in a 10MB slide window across the genome; • Change point copy number - the absolute difference between adjacent segments of the genome; • Segment copy number - the observed absolute copy number state for each segment; • Number of breakpoints per chromosome arm - the number of breaks that occur per chromosome arm; • Length of a segment with oscillating copy number - counting the number of consecutive CN segments that alternate between two copy number states across the genome, rounded to the nearest integer copy number state.

[0119] For each sample, the sum vector of posterior probabilities was calculated using feature component definition as outlined in Macintyre et al. (2018) (see Table 1 above) and the prediction function in the flexmix package in R (Grun et al., 2008).

[0120] This posterior probability sum vector was used to calculate signature exposures using the LCD function in the YAPSA package in Bioconductor (https: / / rdrr.io / bioc / YAPSA / f / README.md) and the signature definition matrix reported by Macintyre et al. (2018) (see Tables 1 and 2). The LCD function in this package can calculate exposures for known signatures, assuming a default signature weight matrix. Classifying samples as either susceptible or resistant to doxorubicin.

[0121] Organoid and spheroid samples: The inventors determined an IC50 threshold to classify samples into those considered sensitive or resistant based on the patient's clinical characteristics. Because the inventors' in vitro drug screening is similar to patients being treated with doxorubicin as monotherapy after a first-line treatment with platinum-based chemotherapy, the inventors estimated that an expected number of sensitive samples based on response would be observed in the clinical trial. Patients resistant to platinum chemotherapy are expected to have an 18% response rate to doxorubicin monotherapy (Mutch et al., 2007; O'Byrne et al., 2002; Kaye et al., 2012; Ferrandina et al., 2008; Gordon et al., 2004; Pujade-Lauraine et al., 2012; Rose et al., 2007), while sensitive patients are expected to have a 28% response rate (Gordon et al., 2001). Patients who relapsed less than 6 months after the first platinum-based chemotherapy were considered resistant, while those who relapsed more than 6 months were considered susceptible (see Table 3). These data allowed us to estimate the expected number of susceptible organoids to be approximately 2 (5 × 0.28 + 5 × 0.18 = 2.3) and the expected number of susceptible spheroids to be 4 (8 × 0.28 + 7 × 0.18 = 3.5). Samples were ranked based on their IC50 and the threshold set that yielded the expected number of susceptible samples in each case.

[0122] Table 3 - Platinum sensitivity status of organoid and spheroid samples [Table 3] Tissue samples: Following treatment with doxorubicin, patients were evaluated for response using clinical symptoms, GCIG CA125 criteria, or CT imaging, and assigned to one of three response categories: complete, stable, or progressive (see Table 4). These categories represent those outlined in the RECIST 1.1 criteria (Rustin et al., 2011).

[0123] Table 4 - Reaction state and prediction for retrospective tissue samples [Table 4]

[0124] Predictive performance evaluation In training (organoids), validation (spheroids), and retrospective (histology) analyses, performance was evaluated using specificity and susceptibility to predict resistance to doxorubicin. Specificity was considered to be the proportion of susceptible and resistant samples correctly identified. A sorting test was performed to assess the significance of observed susceptibility. All possible resistant / susceptible sample labels were calculated, and a predictive threshold yielding 100% specificity based on susceptible sample labels was determined, followed by susceptibility recording. The resulting distribution of all observed susceptibility values ​​was used to determine the probability of observing susceptibility greater than or equal to the observed susceptibility.

[0125] Progression-free survival analysis For each patient, progression-free time was recorded as the number of days from the first dose of doxorubicin to the recorded date of progression, determined by any clinical symptom, GCIG CA125 criteria, or progressive disease on CT imaging. Cox proportional hazards models were fitted in Bioconductor using the survival package (Therneau, 2020) with signature 6 exposure > 0 as a covariate. Signature 6 exposure < 1% was considered = 0. Kaplan-Meier curves were generated in Bioconductor using the survminer package (Kassambara et al., 2019). [Examples]

[0126] Signature 6 exposure predicts resistance to doxorubicin in primary ovarian cancer organoids and in patient-derived spheroids.

[0127] In this embodiment, the inventors aim to test the hypothesis that the characteristics of copy number features indicating the presence of lesion amplification may indicate resistance to micronucleus-inducing chemotherapy using ovarian cancer organoids. Specifically, ten primary ovarian cancer organoids were treated with doxorubicin (in triplicates) on an 8-point log-doubling scale, with the decrease in IC50 in cell viability used as the response measure. Given the clinical characteristics of the patients from whom the organoids were derived, the predicted susceptibility rate to doxorubicin across the organoids was approximately 20%. Therefore, the two organoids with the lowest IC50 were considered susceptible, and the remaining eight organoids were considered resistant (Figure 1). Signature 6 exposure was calculated for each organoid, and a cut point of exposure > 0 was selected to predict resistance in 100% specificity and 100% susceptibility.

[0128] Next, the cutoff defined using organoid data was evaluated in 12 patient-derived spheroids from ascites fluid treated in vitro with doxorubicin as described above. Again, samples with the four lowest IC50s were classified as susceptible, and the remainder as resistant. As shown in Figure 2, samples without exposure to Signature 6 were correctly identified as susceptible, resulting in 100% specificity in this validation data. Two samples that were actually resistant were predicted as susceptible (118947 & 119025), resulting in 82% susceptibility. [Examples]

[0129] Signature 6 exposure predicts resistance to doxorubicin in ovarian cancer patients.

[0130] In this embodiment, the inventors aim to retrospectively validate the study developed in Example 1 for a group of patients treated with doxorubicin. CTCR-OV04 is an established ovarian clinical study to identify biological markers of response. The inventors evaluated Signature 6 exposure in this patient population. After QC and filtering, response data and signatures were available for 24 patients. Signature 6 exposure levels were evaluated for patients in four overall response categories: complete response, partial response (none of the 24 patients in this group were in this category), stable, and progression (Figure 3).

[0131] In this group, using a cutoff of 0 for signature 6 exposure, we identified cases progressing as resistant to doxorubicin with a susceptibility of 0.5 (i.e., 50% of cases actually resistant were predicted to be susceptible, or in other words, 50% of resistant patients were identified) and a specificity of 1 (i.e., all cases predicted to be resistant were actually resistant, or in other words, no cases predicted to be resistant when actually susceptible) (p-value = 0.05). Patients predicted to be resistant were approximately five times more likely to experience disease progression compared to patients predicted to be susceptible, according to the Cox proportional hazards model (Figure 4, HR 5.2, p-value 0.0011, log-rank test).

[0132] In any such test, depending on the cutoff point used, there is a trade-off between the number of false positives and false negatives identified. Given the current state of ovarian cancer care, where the majority of patients receive doxorubicin at some point during a course of treatment, reliably and accurately identifying resistant patients (true negatives, TN) is more important than identifying susceptible patients (true positives, TP). In practice, the inability to provide doxorubicin treatment to patients who might have responded is more undesirable than providing it to some patients who do not respond to it. In other words, any patient who can be exempted from unnecessary treatment is an improvement over maintaining the status quo. Therefore, we choose to apply a cutoff that allows for the identification of resistant patients with 100% accuracy, despite the reduced sensitivity of the test (a larger number of false positives).

[0133] The inventors further aim to verify that samples containing circulating tumor DNA can be used in the methods described herein in a similar manner to samples containing tumor DNA extracted directly from tumor tissue. Therefore, copy number profiles were obtained from cell-free DNA extracted from plasma samples and from DNA extracted from matched tumor tissue for patients for whom sequencing data was available for both types of samples. An example of such copy number analysis for matched samples is shown in Figure 6 (Patient 139), where Figure 6A shows the copy number profiles for tumor tissue and matched ctDNA samples, and Figure 6B shows the corresponding calculated signature exposures. The data demonstrate that very similar copy number profiles can be obtained from ctDNA samples and tumor tissue samples, and therefore, similar signature exposures were obtained. This indicates that copy number features indicating lesion amplification (such as exposure to signature 6) can be calculated from both ctDNA samples and tumor tissue samples. In particular, exposure to Signature 6 was found to be similar between the two types of samples, and both accurately classified patients as resistant (see Table 4). Therefore, both types of samples can be used to predict resistance to doxorubicin. [Examples]

[0134] The copy number characteristics indicating lesion amplification show differences between patients who are resistant to or susceptible to doxorubicin.

[0135] Using data from 18 patients from Example 2 (selected solely due to the availability of the data at the time), we determined whether copy number features associated with components highly weighted in copy number signature 6 could be useful as predictors of resistance to doxorubicin. The behavior of segment copy number features, copy number change point features, and segment size features was investigated for susceptible and resistant samples.

[0136] In particular, for segment copy number features, the following values ​​were calculated for each sample: mean segment copy number, median segment copy number, number of copy number events with a segment copy number greater than 5, proportion of copy number events with a segment copy number greater than 5, sum of posterior probabilities for segment copy number component 8 (cn8, N(30.8672269, 23.15811) - see Table 1), and sum of posterior probabilities for segment copy number component 7 (cn7, N(8.39260927, 3.50149434) - see Table 1). The results are shown in Table 5, where the sign in parentheses next to the patient ID indicates the response status (S=stable, C=complete, P=progressive), and the gray shaded row is a patient predicted to be resistant based on Signature 6 exposure (Example 2). avg seg cn = mean segment copy number, median seg cn = median segment copy number, #events cn>5 = number of copy number events with segment copy numbers greater than 5, %events cn>5 = proportion of copy number events with segment copy numbers greater than 5, sum-of-pos cn8 = sum of posterior probabilities for segment copy number component 8, and sum-of-pos cn7 = sum of posterior probabilities for segment copy number component 7.

[0137] For copy number change point features, the following values ​​were calculated for each sample: mean copy number change point, median copy number change point, number of copy number events with a copy number change point greater than 4, proportion of copy number events with a copy number change point greater than 4, sum of posterior probabilities for copy number change point component 7 (cp7, N(28.7346654,22.0551593) - see Table 1), sum of posterior probabilities for copy number change point component 6 (cp6, N(7.3149416,3.45921997) - see Table 1), and sum of posterior probabilities for copy number change point component 5 (cp5, nN(3.00685766,1.03958107) - see Table 1). The results are shown in Table 6, where the sign in parentheses next to the patient ID indicates the response status (S=stable, C=complete, P=progressive), and the gray shaded row is a patient predicted to be resistant based on Signature 6 exposure (Example 2). avg cp = mean copy number change point, median cp = median copy number change point, #events cp>4 = number of copy number events with copy number change points greater than 4, %events cp>4 = percentage of copy number events with copy number change points greater than 4, sum-of-pos cp7 = sum of posterior probabilities for copy number change point component 7, sum-of-pos cp6 = sum of posterior probabilities for copy number change point component 6, and sum-of-pos cp5 = sum of posterior probabilities for copy number change point component 5.

[0138] For segment size features, for each sample, the following values ​​are obtained: the number of copy number events with a segment size greater than 12,000,000 bp, the proportion of copy number events with a segment size greater than 12,000,000 bp, the number of copy number events with a segment size less than 4,000,000 bp, the proportion of copy number events with a segment size less than 4,000,000 bp, the sum of posterior probabilities for segment size component 7 (ss7, N(16419124.1,5226151.49) - see Table 1), and the sum of posterior probabilities for segment size component 8 (ss8, N( The sum of posterior probabilities for segment size component 9 (29508322.4,9703791.44) - see Table 1) was calculated, as was the sum of posterior probabilities for segment size component 1 (ss1) (N(426861.918,186924.872) - see Table 1), the sum of posterior probabilities for segment size component 2 (ss2) (N(1081858.4,407302.128) - see Table 1), and the sum of posterior probabilities for segment size component 3 (ss3) (N(2233029.82,749092.036) - see Table 1).The results are shown in Table 7, where the sign in parentheses next to the patient ID indicates the response status (S=stable, C=complete, P=progressive), and the gray shaded row is a patient predicted to be resistant based on Signature 6 exposure (Example 2). #events ss>12Mb=number of copy number events with segment sizes greater than 12,000,000 bp, %events ss>12Mb=percentage of copy number events with segment sizes greater than 12,000,000 bp, #events ss<4Mb=number of copy number events with segment sizes less than 4,000,000 bp, %events ss<4Mb=percentage of copy number events with segment sizes less than 4,000,000 bp, sum-of-pos ss7=sum of pos pos ss8=sum of pos pos pos pos ss9 = sum of posterior probabilities for segment size component 9, sum-of-pos ss1 = sum of posterior probabilities for segment size component 1, sum-of-pos ss2 = sum of posterior probabilities for segment size component 2, and sum-of-pos ss3 = sum of posterior probabilities for segment size component 3.

[0139] Table 5 - Characteristics of segment copy number features in tissue sample data [Table 5-1] [Table 5-2]

[0140] The data in Table 5 show that a threshold for mean segment copy number of 2.65–3.60 (or 2.8–3.4, 3–3.2, or approximately 3.15, e.g., 3.146530) can be used to separate samples from resistant vs. susceptible patients, resulting in 5 / 6 of susceptible patients being identified as susceptible and 5 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 2.85–3.1 (or 2.9–3.05, 2.95–3.05, or approximately 3, e.g., 3.000824) for median segment copy number can be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 5 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 7–9 (or 7.5–8.5, 8–9, or approximately 8.5) for the maximum segment copy number may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 7 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 36–60 (or 36–50, 38–48, 40–45, e.g., 42) for events with a segment copy number greater than 5 may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible and 3 / 12 of resistant patients being identified as resistant.

[0141] Alternatively, or in addition to this, a threshold of approximately 5% (or 25-28%) of events with segment copy numbers greater than 5 may be used, resulting in 5 / 6 (and 6 / 6 each) of susceptible patients being identified as susceptible, and 7 / 12 (or 4 / 12 each) of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 7e-5 to 1.1e-4 for the sum of the posterior probabilities of copy number component 8 may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible, and 5 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 1.5e-3 for the sum of the posterior probabilities of copy number component 8 may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible, and 4 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 2e-3 to 6e-3 for the sum of the posterior probabilities of copy number component 8 may be used, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 0.0017 for the sum of the posterior probabilities of copy number component 7 may be used, resulting in 5 / 6 susceptible patients being identified as susceptible and 8 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 0.012 to 0.017 for the sum of the posterior probabilities of copy number component 7 may be used, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 2e-3 to 1.5e-2 (e.g., 0.006882869) for the sum of the posterior probabilities of either copy number components 7 and 8 (i.e., one or both) may be used, resulting in 5 / 6 of susceptible patients (or 6 / 6 if above 1.11e-2) being identified as susceptible, and 6 / 12 of resistant patients (or 5 / 12 if above 3e-3, and 4 / 12 if above 5e-3) being identified as resistant.

[0142] Table 6 - Characteristics of copy number change point features in tissue sample data [Table 6-1] [Table 6-2]

[0143] The data in Table 6 show that a threshold for the mean copy number change point of 1.2–1.328 (or 1.2–1.328, 1.3–1.328, or approximately 1.328, e.g., 1.323483) can be used to separate samples from resistant vs. susceptible patients, resulting in 3 / 6 of susceptible patients being identified as susceptible and 6 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 1.2–2.1 for the median copy number change point can be used, resulting in 3 / 6 of susceptible patients (4 / 6 above 1.7) being identified as susceptible and 5 / 12 of resistant patients (4 / 12 above 1.33) being identified as resistant. Alternatively, or in addition to this, a threshold of 1.04–1.06 (e.g., 1.051904) for the median copy number change point may be used, resulting in 4 / 6 of susceptible patients being identified as susceptible and 6 / 12 of resistant patients being identified as resistant.

[0144] Alternatively, or in addition to this, a threshold of 4.01–5 (or 4.01–4.025) for the maximum copy number change point may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 10 / 12 of resistant patients being identified as resistant (or 9 / 12 above 4.03, and 8 / 12 above 4.28). Alternatively, or in addition to this, a threshold of approximately 18 (e.g., 17.6–18) for the maximum copy number change point may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 18–30 for the maximum copy number change point may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible and 3 / 12 of resistant patients being identified as resistant.

[0145] Alternatively, or in addition to this, a threshold of 2–18 (or 2–8, 2–6, e.g., 3) may be used for events with copy number change points greater than 4, resulting in 5 / 6 of susceptible patients being identified as susceptible (6 / 6 if greater than 11) and 6 / 12 of resistant patients being identified as resistant (4 / 12 if greater than 2, 3 / 12 if greater than 9). Alternatively, or in addition to this, a threshold of 2%–10% (or 2–8, e.g., 5%, or 8.6–10%, e.g., 8.6%) may be used for events with copy number change points greater than 4, resulting in 5 / 6 of susceptible patients being identified as susceptible (6 / 6 if greater than 8.6%) and 4 / 12 of resistant patients being identified as resistant (3 / 12 if greater than 8.6%). Alternatively, or in addition to this, a threshold of 1.91e-5 to 1.1e-3 for the sum of the posterior probabilities of copy number change point component 7 may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible (6 / 6 above 1.034e-3) and 10 / 12 of resistant patients being identified as resistant (9 / 12 above 1.97e-5, 7 / 12 above 3e-5, 4 / 12 above 4e-5, and 4 / 12 between 1.035e-3 and 1.23e-3). Alternatively, or in addition to this, a threshold of 2.8e-4 to 6.6e-3 for the sum of the posterior probabilities of copy number change point component 6 may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible (6 / 6 above 4.72e-3) and 11 / 12 of resistant patients being identified as resistant (10 / 12 above 3.6e-4, 9 / 12 above 6.9e-4, 8 / 12 above 7.3e-4, 7 / 12 above 8.7e-4, 5 / 12 above 1.2e-3, and 4 / 12 between 4.72e-3 and 6.6e-3). Alternatively, or in addition to this, a threshold of 0.006 to 0.023 may be used for the sum of the posterior probabilities of copy number change point component 5, resulting in 5 / 6 of susceptible patients being identified as susceptible and 10 / 12 of resistant patients being identified as resistant (8 / 12 above 0.009, and 5 / 12 above 0.013).Alternatively, or in addition to this, a threshold of 0.02041483 may be used for the sum of the posterior probabilities for any of the change point components in Table 6, resulting in 5 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant.

[0146] Table 7 - Characteristics of segment size features in tissue sample data [Table 7-1] [Table 7-2]

[0147] The data in Table 7 shows that a threshold of 130 for the sum of the number of events with segment sizes greater than 12 Mbp and the number of events with segment sizes less than 4 Mbp can be used to separate samples from resistant vs. susceptible patients, resulting in 5 / 6 of susceptible patients being identified as susceptible and 6 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 52 for the number of events with segment sizes greater than 12 Mbp can be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 8 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 55 for the number of events with segment sizes less than 4 Mbp can be used, resulting in 3 / 6 of susceptible patients being identified as susceptible and 8 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 52 for the number of events with a segment size greater than 12 Mbp may be used in combination with a threshold of 55 for the number of events with a segment size less than 4 Mb, resulting in 5 / 6 of susceptible patients being identified as susceptible and 6 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 0.02 for the sum of the posterior probabilities of segment size component 1 may be used, resulting in 4 / 6 of susceptible patients being identified as susceptible and 7 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.032 for the sum of the posterior probabilities of segment size component 1 may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible and 3 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.032 for the sum of the posterior probabilities of segment size component 2 may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant.Alternatively, or in addition to this, a threshold of 0.033 may be used for the sum of the posterior probabilities of segment size component 2, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 0.03 may be used for the sum of the posterior probabilities of segment size component 3, resulting in 5 / 6 susceptible patients being identified as susceptible and 4 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.033 may be used for the sum of the posterior probabilities of segment size component 3, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant.

[0148] Alternatively, or in addition to this, a threshold of approximately 0.032 for the sum of the posterior probabilities of any of segment size components 1-3 (i.e., a threshold that applies to one, two, or all three) may be used, resulting in 5 / 6 susceptible patients being identified as susceptible and 4 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.033 for the sum of the posterior probabilities of any of segment size components 1-3 (i.e., a threshold that applies to one, two, or all three) may be used, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.014 for the sum of the posterior probabilities of segment size component 7 may be used, resulting in 5 / 6 susceptible patients being identified as susceptible and 10 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.023 for the sum of the posterior probabilities of segment size component 7 may be used, resulting in 6 / 6 susceptible patients being identified as susceptible and 2 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.029 for the sum of the posterior probabilities of segment size component 8 may be used, resulting in 5 / 6 susceptible patients being identified as susceptible and 4 / 12 resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.031 for the sum of the posterior probabilities of segment size component 8 may be used, resulting in 6 / 6 susceptible patients being identified as susceptible and 3 / 12 resistant patients being identified as resistant.

[0149] Alternatively, or in addition to this, a threshold of approximately 0.029 for the sum of the posterior probabilities of any of the segment size components 7-9 (i.e., a threshold that applies to one, two, or all three) may be used, resulting in 5 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.033 for the sum of the posterior probabilities of any of the segment size components 7-9 (i.e., a threshold that applies to one, two, or all three) may be used, resulting in 6 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.02 for the sum of the posterior probabilities of any of segment size components 7–9 (i.e., a threshold that applies to one, two, or all three) may be used in combination with a threshold of approximately 0.03 for the sum of the posterior probabilities of any of segment size components 1–3 (i.e., a threshold that applies to one, two, or all three), resulting in 4 / 6 of susceptible patients being identified as susceptible and 4 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of approximately 0.02041483 for the sum of the posterior probabilities of any combination of segment size components 1–3 and 7–9 in Table 7 (i.e., a threshold that applies to at least one in the entire group) may be used, resulting in 4 / 6 of susceptible patients being identified as susceptible and 8 / 12 of resistant patients being identified as resistant. Alternatively, or in addition to this, a threshold of 0.1527743 may be used for the sum of the posterior probabilities for any of the segment size components in Table 7 (preferably, at least one of ss1-3 and at least one of ss7-9 exceed this threshold).

[0150] In addition to the specific values ​​and ranges of values ​​mentioned above, any threshold that falls within any of the values ​​in Tables 5, 6, or 7, as well as any threshold that, alone or in combination with another threshold that falls within any of the values ​​in Tables 5, 6, or 7, results in 6 / 6 of susceptible patients being identified as susceptible and at least one of the resistant patients being identified as resistant, are explicitly recalled. Indeed, any such threshold or combination of thresholds results in at least one patient being exempted from ineffective treatments, without any susceptible patients being excluded from potentially effective treatments. Furthermore, any equivalent threshold or combination of thresholds having these properties, and determined using different training datasets (e.g., different groups of patient-derived copy number features associated with known or assumed resistance status), may also be used.

[0151] References 1.Macintyre G, Goranova TE, De Silva D, et al.Copy number signatures and mutational processes in ovarian carcinoma.Nat Genet.2018;50(9):1262-1270.doi:10.1038 / s41588-018-0179-8 2. Alexandrov L, Kim J, Haradhvala NJ, Huang MN, Ng AWT, Boot A, et al. The Repertoire of Mutational Signatures in Human Cancer. bioRxiv2018:322859 doi10.1101 / 322859. 3.Schulze K,Imbeaud S,Letouze E,Alexandrov LB,Calderaro J,Rebouissou S,et al.Exome sequencing of hepatocellular carcinomas identifies new mutational signatures and potential therapeutic targets.Nature Genetics2015;47:505 doi10.1038 / ng.3252. 4.Alexandrov LB,Nik-Zainal S,Wedge DC,Aparicio SAJR,Behjati S,Biankin AV,et al.Signatures of mutational processes in human cancer.Nature2013;500(7463):415-21doi10.1038 / nature12477. 5.Mutch,D.G.et al.Randomized phase III trial of gemcitabine compared with pegylated liposomal doxorubicin in patients with platinum-resistant ovarian cancer.J.Clin.Oncol.25,2811-2818(2007). 6.O’Byrne,K.J.,Bliss,T.V.P.&Graham,J.D.A Phase III study of Doxil / Caylex versus paclitaxel in platinum treated taxane naive relapsed ovarian cancer.(2002). 7.Kaye,S.B.et al.Phase II,open-label,randomized,multicenter study comparing the efficacy and safety of olaparib,a poly(ADP-ribose)polymerase inhibitor,and pegylated liposomal doxorubicin in patients with BRCA1 or BRCA2 mutations and recurrent ovarian cancer.J.Clin. Oncol.30,372-379(2012). 8.Ferrandina,G.et al.Phase III trial of gemcitabine compared with pegylated liposomal doxorubicin in progressive or recurrent ovarian cancer.J.Clin.Oncol.26,890-896(2008). 9.Gordon,A.N.,Tonda,M.,Sun,S.,Rackoff,W.&Doxil Study30-49 Investigators.Long-term survival advantage for women treated with pegylated liposomal doxorubicin compared with topotecan in a phase3 randomized study of recurrent and refractory epithelial ovarian cancer.Gynecol.Oncol.95,1-8(2004). 10.Pujade-Lauraine,E.et al.AURELIA:A randomized phase III trial evaluating bevacizumab(BEV)plus chemotherapy(CT)for platinum(PT)-resistant recurrent ovarian cancer(OC).J.Clin.Orthod.30, LBA5002-LBA5002(2012). 11.Rose,P.et al.Phase3 Study:Canfosfamide(C, TLK286)plus carboplatin(P)vs liposomal doxorubicin(D)as2nd line therapy of platinum(P)resistant ovarian cancer(OC).J.Clin.Orthod.25,LBA5529-LBA5529(2007). 12.Bafaloukos,D.et al.A randomized phase II study of carboplatin plus pegylated liposomal doxorubicin versus carboplatin plus paclitaxel in platinum sensitive ovarian cancer patients: a Hellenic Cooperative Oncology Group study.BMC Med.8,3(2010). 13.Alberts,D.S.et al.Randomized trial of pegylated liposomal doxorubicin(PLD)plus carboplatin versus carboplatin in platinum-sensitive(PS)patients with recurrent epithelial ovarian or peritoneal carcinoma after failure of initial platinum-based chemotherapy(Southwest Oncology Group Protocol S0200).Gynecol. Oncol.108,90-94(2008). 14.Ritz C,Baty F,Streibig JC, Gerhard D.Dose-Response Analysis Using R.PLoS One2015;10:e0146021. 15.Li H,Durbin R.Fast and accurate short read alignment with Burrows-Wheeler transform.Bioinformatics2009;25:1754-1760. 16.Broad Institute.Picard toolkit.2018.http: / / broadinstitute.github.io / picard / . 17.Scheinin I,Sie D,Bengtsson H,van de Wiel MA,Olshen AB,van Thuijl HF et al.DNA copy number analysis of fresh and formalin-fixed specimens by shallow whole-genome sequencing with identification and exclusion of problematic regions in the genome assembly.Genome Res2014;24:2022-2032. 18.Grun B,Leisch F.FlexMix Version2:Finite Mixtures with Concomitant Variables and Varying and Constant Parameters.Journal of Statistical Software,Articles2008;28:1-35. 19.Gordon AN,Fleagle JT,Guthrie D,Parkin DE, Gore ME,Lacave AJ.Recurrent epithelial ovarian carcinoma: a randomized phase III study of pegylated liposomal doxorubicin versus topotecan.J Clin Oncol2001;19:3312-3322. 20.Rustin GJS, Vergote I, Eisenhauer E, Pujade-Lauraine E, Quinn M, Thigpen T et al.Definitions for response and progression in ovarian cancer clinical trials incorporating RECIST1.1 and CA125 agreed by the Gynecological Cancer Intergroup (GCIG).Int J Gynecol Cancer2011;21:419-423. 21.Therneau TM.A Package for Survival Analysis in R.2020 https: / / CRAN.R-project.org / package=survival. 22. Kassambara A, Kosinski M, Biecek P. survminer:Drawing Survival Curves using'ggplot2'.2019.https: / / CRAN.R-project.org / package=survminer.

[0152] All references listed herein are incorporated herein by reference, in their entirety, to the same extent as each individual publication or patent or patent application is specifically and individually incorporated by reference.

[0153] The specific embodiments described herein are provided as examples and not as limitations. Any subtitles within this specification are included for convenience only and should not be construed in any way as limiting this disclosure.

Claims

1. A method for predicting the therapeutic response to topoisomerase II toxin in cancer patients, (a) A step of obtaining a tumor copy number profile for the patient, (b) A step of analyzing the tumor copy number profile to evaluate whether the characteristics of one or more copy number features indicate the presence of lesion amplification in the tumor genome, The one or more copy number features are selected from a copy number change point, segment size, and segment copy number, and the one or more copy number features include at least the copy number change point. The step of analyzing the tumor copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, i. For each copy number event in the tumor copy number profile, quantify the one or more copy number features described above, ii. Obtaining one or more aggregation measures for each quantified copy number feature, (c) If the characteristic in at least one copy number feature indicates the presence of lesion amplification in the tumor genome, the patient is likely to be resistant to treatment with the topoisomerase II toxin, Equipped with, (i) The method includes the steps of obtaining an aggregation scale that captures the contribution of each of the one or more copy number features, and comparing the aggregation scale to a predetermined threshold, wherein the aggregation scale exceeds the predetermined threshold, indicating the presence of lesion amplification. The aggregation metric is exposure (Ei) to copy number signature i (SbCi), where Ei is given by the following formula: PbC≒E×SbC (Formula 1) It is a value that satisfies the condition, and in the formula, E is a vector of size n containing coefficients E1, ..., n, and Ei is the exposure to signature i. PbC has a size c ≥ 1, and each element in the vector represents the sum of the posterior probabilities of each copy number event in the tumor copy number profile, where each component C is a distribution of values ​​for the copy number features. The aforementioned one or more components include at least a component C1 which is a distribution of copy number change point values ​​centered around values ​​2 to 32, SbC is a matrix of size c × n, where each value represents the weight of component C in signature i. (ii) The one or more aggregation measures are The mean or median of the distribution of copy number change points over copy number events, The maximum value of the copy number change point over the copy number event, The percentage of copy count events that are above the value of the copy count change point exceeds a further predetermined threshold, The number or percentage of copy number events in which the copy number change point exceeds a further predetermined threshold, and The sum of the previous and subsequent probabilities of the copy number change points for each copy number event belonging to one or more predetermined distributions. A method that includes an aggregated scale selected from the following.

2. The method according to claim 1, wherein the patient is diagnosed with or is likely to have cancer with high chromosomal instability.

3. The method according to claim 1 or 2, wherein the patient is a patient diagnosed with an epithelial malignant tumor or sarcoma.

4. The method according to any one of claims 1 to 3, wherein the patient is diagnosed with glioblastoma, lung cancer, esophageal cancer, pancreatic cancer, breast cancer, or ovarian cancer.

5. The method according to claim 4, wherein the patient has high-grade serous ovarian cancer (HGSOC) or triple-negative breast cancer.

6. The method according to any one of claims 1 to 5, wherein the topoisomerase II toxin is an anthracycline.

7. The method according to claim 6, wherein the topoisomerase II toxin is doxorubicin.

8. The method according to claim 7, wherein the topoisomerase II toxin is liposomal doxorubicin.

9. The method according to any one of claims 1 to 8, wherein the tumor copy number profile for the patient is obtained from a tumor sample from the patient or a liquid biopsy sample from the patient.

10. The step of analyzing the tumor copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, i. For each copy number event in the tumor copy number profile, one or more further copy number features selected from the number of breakpoints per xMB, the number of breakpoints per chromosome arm, and the length of the segment having an oscillating copy number, ii. The method according to any one of claims 1 to 9, comprising obtaining one or more aggregation measures for each further digitized copy number feature.

11. The method according to claim 10, wherein x is 10 in the number of breakpoints per x MB.

12. The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, The method according to any one of claims 1 to 11, comprising comparing the mean, median, or maximum copy number change point with a predetermined threshold, wherein a comparison of the mean, median, or maximum copy number change point exceeding the predetermined threshold indicates the presence of lesion amplification in the tumor genome, and the predetermined threshold is determined by comparing the mean or median copy number change points between resistant and susceptible samples in a training group.

13. The step of analyzing the copy number profile to assess whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome, The method according to any one of claims 1 to 11, comprising comparing the number or proportion of events having copy number change points above a first default threshold with a second default threshold, wherein the number or proportion of events above the second default threshold indicates the presence of lesion amplification in the tumor genome, and the first and / or second default thresholds are obtained by comparing the distribution of copy number change points for resistant and susceptible samples within a training group.

14. The step of analyzing the tumor copy number profile to evaluate whether the characteristics of at least one copy number feature indicate the presence of lesion amplification in the tumor genome is, This includes comparing the sum of the posterior probabilities of the copy number change point values ​​for each copy number event belonging to one or more predetermined distributions with the respective predetermined thresholds, The one or more predetermined distributions are This is the distribution of copy number change point values ​​(C1) centered around values ​​between 2 and 32. The method according to any one of claims 1 to 11, wherein each of the aforementioned default thresholds is quantified by comparing the sum of the posterior probabilities for tolerable and susceptible samples within the training group.

15. The method according to claim 14, wherein the distribution of copy number change point values ​​(C1) is centered around values ​​of 4 to 30, 5 to 30, or 6 to 30.

16. The method according to any one of claims 1 to 15, wherein PbC is a vector of size 1 ≤ c ≤ 36.

17. The method according to any one of claims 1 to 16, wherein the one or more components include at least one component C2 which is a distribution of copy values ​​centered around values ​​5 to 34.

18. The method according to any one of claims 1 to 17, wherein the one or more components include at least one component C3 having a segment size distribution centered around a value of 100,000 to 4,000,000 base pairs, and at least one component C4 having a segment size distribution centered around a value of 12,000,000 to 80,000,000 bp.

19. The method according to claim 18, wherein the one or more components include components 1 to 36 in Table 1, or a distribution having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1.

20. The method according to claim 19, wherein the SbC element for each component is the weight specified in Table 1, or a weight within 10%, 5%, 2%, or 1% of the weights in Table 1.

21. The method according to claim 20, wherein the one or more components include components 1 to 36 in Table 1, and the elements of SbC are the weights in Table 1.

22. The method according to any one of claims 1 to 21, wherein the one or more components include components 1 to 36 in Table 1, or a distribution having a mean (or λ) and / or standard deviation within 10%, 5%, 2%, or 1% of the distribution parameters in Table 1, and the elements of SbCi for each component are the weights defined in Table 1, or weights within 10%, 5%, 2%, or 1% of the weights in Table 1, and SbC further includes weights for additional signatures defined in Table 2, or weights within 10%, 5%, 2%, or 1% of the weights in Table 2.

23. The method according to claim 22, wherein exposure to signature i is calculated using all seven signatures in Tables 1 and 2, or the corresponding signatures, and the exposure Ei indicating the presence of lesion amplification is weight signature 6 in the signatures in Table 1.

24. A system for predicting the treatment response of cancer patients, At least one processor, A system comprising at least one non-temporary computer-readable medium containing instructions, wherein, when executed by the at least one processor, the instructions cause the at least one processor to perform each step of the method according to any one of claims 1 to 23.

Images