|
|
|
|
||
Building IP: JUNO Patent Appl "METHODS OF IDENTIFYING FEATURES ASSOCIATED WITH...METHODS OF IDENTIFYING FEATURES ASSOCIATED WITH CLINICAL RESPONSE AND USES THEREOFDOCUMENT IDUS 20230178239 A1 DATE PUBLISHED2023-06-08 INVENTOR INFORMATION NAMECITYSTATEZIP CODECOUNTRYHAUSE, JR.; Ronald James Seattle WA N/A US JIANG; Yue Seattle WA N/A US APPLICANT INFORMATION NAME Juno Therapeutics, Inc. CITY Seattle STATE WA ZIP CODE N/A COUNTRY US AUTHORITY N/A TYPE assignee ASSIGNEE INFORMATION NAME Juno Therapeutics, Inc. CITY Seattle STATE WA ZIP CODE N/A COUNTRY US TYPE CODE 02 APPLICATION NO17/924652 DATE FILED2021-05-12 DOMESTIC PRIORITY (CONTINUITY DATA)us-provisional-application US 63037592 20200610 us-provisional-application US 63024494 20200513 AbstractThe present disclosure relates to methods for identifying features, such as attributes of subjects, therapeutic cell compositions, and input compositions used to produce therapeutic cell compositions, associated with clinical responses of subjects, e.g., patients, following treatment with the therapeutic cell composition in connection with a cell therapy. The cells of the therapeutic cell composition express recombinant receptors such as chimeric receptors, e.g., chimeric antigen receptors (CARs) or other transgenic receptors such as T cell receptors (TCRs). The methods provide for the identification of features associated with clinical responses. In some embodiments, the methods can be used to determine (e.g., predict) a subject's response to treatment with the therapeutic cell composition. Background/SummaryCROSS-REFERENCE TO RELATED APPLICATIONS [0001] This application claims priority from U.S. provisional application No. 63/024,494, filed May 13, 2020, entitled “METHODS OF IDENTIFYING FEATURES ASSOCIATED WITH CLINICAL RESPONSE AND USES THEREOF,” and U.S. provisional application No. 63/037,592, filed Jun. 10, 2020, entitled “METHODS OF IDENTIFYING FEATURES ASSOCIATED WITH CLINICAL RESPONSE AND USES THEREOF,” the contents of which are incorporated by reference in their entirety. INCORPORATION BY REFERENCE OF SEQUENCE LISTING [0002] The present application is being filed along with a Sequence Listing in electronic format. The Sequence Listing is provided as a file entitled 735042023740SeqList.TXT, created May 8, 2021, which is 51,373 bytes in size. The information in the electronic format of the Sequence Listing is incorporated by reference in its entirety. FIELD [0003] The present disclosure relates to methods for identifying features, such as attributes of subjects, therapeutic cell compositions, and input compositions used to produce therapeutic cell compositions, associated with clinical responses of subjects, e.g., patients, following treatment with the therapeutic cell composition in connection with a cell therapy. The cells of the therapeutic cell composition express recombinant receptors such as chimeric receptors, e.g., chimeric antigen receptors (CARs) or other transgenic receptors such as T cell receptors (TCRs). The methods provide for the identification of features associated with clinical responses. In some embodiments, the methods can be used to determine (e.g., predict) a subject's response to treatment with the therapeutic cell composition. BACKGROUND [0004] Various immunotherapy and/or cell therapy methods are available for treating diseases and conditions. For example, adoptive cell therapies (including those involving the administration of cells expressing chimeric receptors specific for a disease or disorder of interest, such as chimeric antigen receptors (CARs) and/or other recombinant antigen receptors, as well as other adoptive immune cell and adoptive T cell therapies) can be beneficial in the treatment of cancer or other diseases or disorders. Improved approaches are needed for determining whether a treatment will result in a beneficial clinical response. Provided herein are methods that address such needs. SUMMARY [0005] Provided herein are methods of identifying features associated with a clinical response, the method comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a CAR; and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more of the subject features, one or more the input composition features, and one or more of the therapeutic cell composition features; (c) obtaining clinical responses from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features and the obtained clinical responses from the plurality of subjects as input to train a random forests model using supervised learning; and (e) identifying from the trained random forests model the informative features associated with the clinical responses. [0006] Provided herein are methods of identifying features associated with a clinical response, the method comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more of the subject features, one or more of the input composition features, and one or more of the therapeutic cell composition features; (c) obtaining clinical responses over time from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features and clinical responses from the plurality of subjects as input to train a random survival forests model using supervised learning; and (e) identifying from the trained random survival forests model the informative features associated with the clinical responses. [0007] In some embodiments of any of the methods provided herein, the identifying the informative features associated with the clinical responses comprises determining an importance measure for each of the informative features. In some embodiments, the importance measure comprises a permutation importance measure, a mean minimal depth, and/or a total number of trees from the random forests, e.g., the trained random forests model, wherein the informative feature splits a root node. In some embodiments, the importance measure comprises a permutation importance measure, a mean minimal depth, and/or a total number of trees from the random survival forests, e.g., the trained random survival forests model, wherein the informative feature splits a root node. In some embodiments, the importance measure is the permutation importance measure. In some embodiments, the importance measure is the mean minimal depth. In some embodiments, the importance measure is the total number of trees from the random forests, e.g., the trained random forests model, wherein the informative feature splits a root node. In some embodiments, the importance measure is the total number of trees from the random survival forests, e.g., the trained random survival forests model, wherein the informative feature splits a root node. In some embodiments, the informative features associated with the clinical responses are the first 10, 9, 8, 7, 6, 5, 4, 3, 2, or 1 informative features identified by rank ordering, e.g., rank ordering values, of the importance measure for each of the informative features, wherein the importance measure is the same for each informative feature. In some embodiments, the informative features associated with the clinical responses are the first 5 informative features identified by rank ordering values of the importance measure for each of the informative features, wherein the importance measure is the same for each informative feature. In some embodiments, the informative features associated with the clinical responses is the first informative feature identified by rank ordering values of the importance measure for each of the informative features, wherein the importance measure is the same for each informative feature. [0008] Provided herein are methods of determining, e.g., predicting, a clinical response, the method comprising: (a) receiving features comprising: (i) subject features determined from a subject prior to the subject being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from an input composition, wherein the input composition comprises T cells selected from a sample from the subject, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from the therapeutic cell composition, wherein the therapeutic cell composition is produced from the input composition and expresses the CAR, wherein the therapeutic composition is to be administered to the subject; and (b) applying the features as input to a random forests model trained to determine, e.g., predict, based on informative features identified by preprocessing, clinical responses of the subject to treatment with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random forests model. [0009] Provided herein are methods of determining, e.g., predicting, a clinical response, the method comprising: (a) receiving features comprising: (i) subject features determined from a subject prior to the subject being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from an input composition, wherein the input composition comprises T cells selected from a sample from the subject, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from the therapeutic cell composition, wherein the therapeutic cell composition is produced from the input composition and expresses the CAR, wherein the therapeutic composition is to be administered to the subject; and (b) applying the features as input to a random survival forests model trained to determine, e.g., predict, based on informative features identified by preprocessing, clinical responses of the subject to treatment with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random survival forests model. [0010] Provided herein are methods of treating a subject, the method comprising: (a) selecting T cells from a sample from a subject to produce an input composition comprising T cells; (b) determining features comprising: (i) subject features determined from a subject prior to the subject being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from an input composition, wherein the input composition comprises T cells selected from a sample from the subject, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from the therapeutic cell composition, wherein the therapeutic cell composition is produced from the input composition and expresses the CAR, wherein the therapeutic composition is to be administered to the subject; and (c) applying the features as input to a random forests model trained to determine, e.g., predict, based on informative features identified by preprocessing, clinical responses of the subject to treatment with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random forests model; and administering a treatment to the subject wherein: (1) if the subject is determined, e.g., predicted, to have a clinical response selected from the group consisting of a complete response (CR), a partial response (PR), a durable response of greater than 3 months, progression free survival (PFS) for more than 3 months, overall response rate (ORR), objective response (OR), a desired pharmacokinetic response that is or is greater than a target pharmacokinetic response, and no or a mild toxicity response (optionally wherein the toxicity is grade 2 or less CRS or grade 2 or less neurotoxicity), a predetermined treatment regimen comprising the therapeutic cell composition is administered; or (2) if the subject is determined, e.g., predicted, to have a clinical response selected from the group consisting of a toxicity response (optionally wherein the toxicity response is a severe cytokine release syndrome or severe neurotoxicity), a reduced pharmacokinetics response compared to a target pharmacokinetic response, progressive disease (PD), a durable response of less than 3 months, and PFS of less than 3 months, administering to the subject a treatment regimen comprising the therapeutic cell composition that is altered compared to the predetermined treatment regimen comprising the therapeutic cell composition. [0011] Provided herein are methods of treating a subject, the method comprising: (a) selecting T cells from a sample from a subject to produce an input composition comprising T cells; (b) determining features comprising: (i) subject features determined from a subject prior to the subject being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from an input composition, wherein the input composition comprises T cells selected from a sample from the subject, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from the therapeutic cell composition, wherein the therapeutic cell composition is produced from the input composition and expresses the CAR, wherein the therapeutic composition is to be administered to the subject; and (c) applying the features as input to a random survival forests model trained to determine, e.g., predict, based on informative features identified by preprocessing, clinical responses in the subject to be treated with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random survival forests model; and administering a treatment to the subject wherein: (1) if the subject is determined, e.g., predicted, to have a clinical response selected from the group consisting of a complete response (CR), a partial response (PR), a durable response of greater than 3 months, progression free survival (PFS) for more than 3 months, overall response rate (ORR), objective response (OR), a desired pharmacokinetic response that is or is greater than a target pharmacokinetic response, and no or a mild toxicity response (optionally wherein the toxicity is grade 2 or less CRS or grade 2 or less neurotoxicity), a predetermined treatment regimen comprising the therapeutic cell composition is administered; or (2) if the subject is determined, e.g., predicted, to have a clinical response selected from the group consisting of a toxicity response (optionally wherein the toxicity response is a severe cytokine release syndrome or severe neurotoxicity), a reduced pharmacokinetics response compared to a target pharmacokinetic response, progressive disease (PD), a durable response of less than 3 months, and PFS of less than 3 months, administering to the subject a treatment regimen comprising the therapeutic cell composition that is altered compared to the predetermined treatment regimen comprising the therapeutic cell composition. [0012] In some of any embodiments, the method further comprises generating the therapeutic cell composition. [0013] Provided herein in some embodiments is a method of treating a subject, the method comprising: (a) selecting T cells from a sample from a subject to produce an input composition comprising T cells; (b) generating a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition (i) is for treating the disease or condition, (ii) is produced from the input composition, and (iii) is to be administered to the subject; (c) determining features comprising: (i) subject features determined from the subject prior to the subject being treated with the therapeutic cell composition; (ii) input composition features determined from the input composition; and (iii) therapeutic cell composition features determined from the therapeutic cell composition; (c) applying the features as input to a random forests model trained to determine, based on informative features identified by preprocessing, a clinical response in the subject to be treated with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random forests model; and (d) administering a treatment to the subject wherein: (1) if the subject is determined to have a clinical response selected from the group consisting of a complete response (CR), a partial response (PR), a durable response of greater than 3 months, progression free survival (PFS) for more than 3 months, objective response (OR), a desired pharmacokinetic response that is or is greater than a target pharmacokinetic response, and no or a mild toxicity response (optionally wherein the mild toxicity response is grade 2 or less cytokine release syndrome (CRS) or grade 2 or less neurotoxicity), a predetermined treatment regimen comprising the therapeutic cell composition is administered; or (2) if the subject is determined to have a clinical response selected from the group consisting of a toxicity response (optionally wherein the toxicity response is a severe cytokine release syndrome (CRS) or severe neurotoxicity), a reduced pharmacokinetics response compared to a target pharmacokinetic response, progressive disease (PD), a durable response of less than 3 months, and PFS of less than 3 months, administering to the subject a treatment regimen comprising the therapeutic cell composition that is altered compared to the predetermined treatment regimen comprising the therapeutic cell composition. [0014] Provided herein in some embodiments is a method of treating a subject, the method comprising: (a) selecting T cells from a sample from a subject to produce an input composition comprising T cells; (b) generating a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition (i) is for treating the disease or condition, (ii) is produced from the input composition, and (iii) is to be administered to the subject; (c) determining features comprising: (i) subject features determined from the subject prior to the subject being treated with the therapeutic cell composition; (ii) input composition features determined from the input composition; and (iii) therapeutic cell composition features determined from the therapeutic cell composition; (c) applying the features as input to a random survival forests model trained to determine based on informative features identified by preprocessing, a clinical response in the subject to be treated with the therapeutic cell composition prior to treating the subject with the therapeutic cell composition, wherein the features applied as input are the same informative features as those used to train the random survival forests model; and (d) administering a treatment to the subject wherein: (1) if the subject is determined to have a clinical response selected from the group consisting of a complete response (CR), a partial response (PR), a durable response of greater than 3 months, progression free survival (PFS) for more than 3 months, objective response (OR), a desired pharmacokinetic response that is or is greater than a target pharmacokinetic response, and no or a mild toxicity response (optionally wherein the mild toxicity response is grade 2 or less cytokine release syndrome (CRS) or grade 2 or less neurotoxicity), a predetermined treatment regimen comprising the therapeutic cell composition is administered; or (2) if the subject is determined to have a clinical response selected from the group consisting of a toxicity response (optionally wherein the toxicity response is a severe cytokine release syndrome (CRS) or severe neurotoxicity), a reduced pharmacokinetics response compared to a target pharmacokinetic response, progressive disease (PD), a durable response of less than 3 months, and PFS of less than 3 months, administering to the subject a treatment regimen comprising the therapeutic cell composition that is altered compared to the predetermined treatment regimen comprising the therapeutic cell composition. [0015] In some embodiments, the random forests model is trained to determine if the subject will have a complete response (CR). In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have a complete response (CR), or (2) the subject is administered the altered treatment regimen if the subject is determined to have progressive disease (PD). [0016] In some embodiments, the random forests model is trained to determine if the subject will have a partial response (PR). In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have a partial response (PR), or (2) the subject is administered the altered treatment regimen if the subject is determined to have progressive disease (PD). [0017] In some embodiments, the random forests model is trained to determine if the subject will have a durable response of greater than 3 months. In some embodiments, the random forests survival model is trained to determine if the subject will have a durable response of greater than 3 months. In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have a durable response of greater than three months; or (2) the subject is administered the altered treatment regimen if the subject is determined to have a durable response of less than three months. [0018] In some embodiments, the random forests model is trained to determine if the subject will have progression free survival (PFS) for more than 3 months. In some embodiments, the random survival forests model is trained to determine if the subject will have progression free survival (PFS) for more than 3 months. In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have progression free survival (PFS) for more than three months, or (2) the subject is administered the altered treatment regimen if the subject is determined to have progression free survival (PFS) of less than three months. [0019] In some embodiments, the random forests model is trained to determine if the subject will have an objective response (OR). In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have an objective response (OR), or (2) the subject is administered the altered treatment regimen if the subject is determined to have progressive disease (PD). [0020] In some embodiments, the random forests model is trained to determine a pharmacokinetic response of the subject. In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have a desired pharmacokinetic response that is or is greater than a target pharmacokinetic response, or (2) the subject is administered the altered treatment regimen if the subject is determined to have a reduced pharmacokinetic response compared to the target pharmacokinetic response. [0021] In some embodiments, the random forests model is trained to determine if the subject will have a toxicity response. In some embodiments, (1) the subject is administered the predetermined treatment regimen if the subject is determined to have no or a mild toxicity response, or (2) the subject is administered the altered treatment regimen if the subject is determined to have a toxicity response. In some embodiments, the toxicity response is severe CRS. In some embodiments, the toxicity response is severe neurotoxicity. [0022] In some embodiments, the random forests model is trained using supervised training, the supervised training comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more subject features, one or more input composition features, and one or more therapeutic cell composition features; (c) obtaining clinical responses from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features from a plurality of subjects and the obtained clinical responses as input to train a random forests model. [0023] In some embodiments, the random survival forests model is trained using supervised training, the supervised training comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more subject features, one or more input composition features, and one or more therapeutic cell composition features; (c) obtaining clinical responses over time from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features and clinical responses from the plurality of subjects as input to train a random survival forests model using supervised learning. [0024] Provided herein are methods of developing a random forests model comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more subject features, one or more input composition features, and one or more therapeutic cell composition features; (c) obtaining clinical responses from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features from a plurality of subjects and the obtained clinical responses as input to train a random forests model. [0025] Provided herein are methods of developing a random survival forests model comprising: (a) receiving features comprising: (i) subject features determined from each of a plurality of subjects prior to the subjects being treated with a therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR) that binds to an antigen associated with a disease or condition, wherein the therapeutic cell composition is for treating the disease or condition; (ii) input composition features determined from each of a plurality of input compositions, wherein each of the plurality of input compositions comprises T cells selected from a sample from each of the plurality of subjects, wherein the T cells are used for producing the therapeutic cell composition comprising T cells comprising a chimeric antigen receptor (CAR); and (iii) therapeutic cell composition features determined from each of a plurality of therapeutic cell compositions, wherein each of the plurality of therapeutic cell compositions is produced from one of the plurality of input compositions and expresses the CAR, wherein the therapeutic composition is to be administered to one of the plurality of subjects; (b) preprocessing the features to identify informative features, the informative features comprising a subset of the features comprising one or more subject features, one or more input composition features, and one or more therapeutic cell composition features; (c) obtaining clinical responses over time from each of the plurality of subjects following treatment with one of the plurality of therapeutic compositions; (d) applying the informative features and clinical responses from the plurality of subjects as input to train a random survival forests model using supervised learning. [0026] In some embodiments, each of the plurality of subjects is administered one of the plurality of therapeutic cell compositions, wherein the one therapeutic cell composition administered to the subject is the therapeutic cell composition produced from the input composition of the sample from the subject. [0027] In some embodiments, the preprocessing to identify informative features comprises one or more of: a) removing subject features, input composition features, and therapeutic cell composition features having greater than, than about, or 50% of the data missing; b) removing subject features, input composition features, and therapeutic cell composition features having zero variance or greater than, than about or 95% of data values equal to a single value and/or fewer than 0.1n unique values, wherein n=number of samples; c) imputing missing data for subject features, input composition features, and therapeutic cell composition features by multivariate imputation by chained equations; d) identifying covariate clusters, the covariate clusters comprising sets of subject features, input composition features, and therapeutic cell composition features and combinations thereof, with correlation coefficients of greater than, about, or equal to 0.5, and iteratively selecting subject features, input composition features, and therapeutic cell composition features from the covariate cluster, wherein the selected subject features, input composition features, and therapeutic cell composition features have the lowest mean absolute correlation with all remaining subject features, input composition features, and therapeutic cell composition features. In some embodiments, the preprocessing to identify informative features comprises or is removing subject features, input composition features, and therapeutic cell composition features having greater than, about, or 50% of the data is missing. In some embodiments, the preprocessing to identify informative features comprises or is removing subject features, input composition features, and therapeutic cell composition features having zero variance or greater than, about or 95% of data values equal to a single value and fewer than 0.1n unique values, wherein n=number of samples. In some embodiments, the preprocessing to identify informative features comprises or is imputing missing data for subject features, input composition features, and therapeutic cell composition features by multivariate imputation by chained equations. In some embodiments, the preprocessing to identify informative features comprises or is identifying covariate clusters, the covariate clusters comprising sets of subject features, input composition features, and therapeutic cell composition features and combinations thereof with correlation coefficients of greater than, about, or equal to 0.5, and iteratively selecting subject features, input composition features, and therapeutic cell composition features from the covariate cluster, wherein the selected subject features, input composition features, and therapeutic cell composition features have the lowest mean absolute correlation with all remaining subject features, input composition features, and therapeutic cell composition features. [0028] In some embodiments, the random forests model is evaluated using cross validation. In some embodiments, the random survival forests model is evaluated using cross validation. In some embodiments, the cross validation is or is at least 10-fold cross validation. In some embodiments, the cross validation is nested cross validation. [0029] In some embodiments, the plurality of subjects is or is about 500, 400, 300, 200, 150, 100, 50, 25, 15, or 10 subjects, or is any number between any of the foregoing. In some embodiments, the plurality of subjects is, is about, or is greater than 10 subjects and less than 250 subjects. In some embodiments, the plurality of subjects is, is about, or is greater than 20 subject and less than 200 subjects. In some embodiments, the plurality of subjects is, is about, or is greater than 20 and less than 150 subjects. In some embodiments, the plurality of subjects is, is about, or is greater than 20 subjects and less than 150 subjects. In some embodiments, the plurality of subjects is, is about, or is greater than 20 subjects and less than 100 subjects. In some embodiments, the plurality of subjects are participating in a clinical trial. [0030] In some embodiments, the subject features comprise one or more of subject attributes and clinical attributes. In some embodiments, the subject attributes comprise one or more of age, weight, height, ethnicity, race, sex, and body mass index. In some embodiments, the clinical attributes comprise one or more of biomarkers, disease diagnosis, disease burden, disease duration, disease grade, and treatment history. In some embodiments, the input composition features comprise cell phenotypes. In some embodiments, the therapeutic cell composition features comprise one or more of a cell phenotype, a recombinant receptor-dependent activity, and a dose. In some embodiments, the clinical responses comprise one or more of a complete response (CR), a partial response (PR), a durable response, progression free survival (PFS), overall response rate (ORR), objective response (OR), a pharmacokinetic response that is or is greater than a target pharmacokinetic response, no or a mild toxicity response, a toxicity response, a reduced pharmacokinetics response compared to a target response, or a lack of CR, PR, durable response, ORR, OR, or PFS. [0031] In some embodiments, the clinical response is or comprises a complete response (CR), a partial response (PR), a durable response, progression free survival (PFS), objective response (OR), a pharmacokinetic response that is or is greater than a target pharmacokinetic response, no or a mild toxicity response, a toxicity response, a reduced pharmacokinetics response compared to a target response, or a lack of CR, PR, durable response, or objective response (OR). [0032] In some embodiments, the clinical response is a complete response (CR). In some embodiments, the clinical response is a lack of complete response (CR). In some embodiments, the clinical response is a partial response (PR). In some embodiments, the clinical response is a lack of partial response (PR). In some embodiments, the clinical response is an objective response (OR). In some embodiments, the clinical response is a lack of objective response (OR). In some embodiments, the clinical response is a toxicity response. In some embodiments, the clinical response is a lack of toxicity response. In some embodiments, the toxicity response is a mild toxicity response. In some embodiments, the toxicity response is a severe toxicity response. In some embodiments, the toxicity response is severe CRS. In some embodiments, the toxicity response is severe neurotoxicity. In some embodiments, the clinical response is a durable response. In some embodiments, the clinical response is a lack of durable response. In some embodiments, the clinical response is the duration of response (DOR). In some embodiments, the clinical response is a duration of response (DOR) of at least or at least about three months. In some embodiments, the clinical response is progression free survival (PFS). In some embodiments, the clinical response is progression free survival (PFS) of at least or at least about three months. In some embodiments, the clinical response is a pharmacokinetic response that is or is greater than a target pharmacokinetic response. In some embodiments, the pharmacokinetic response is a measure of expansion of CAR T cells of the therapeutic cell composition following treatment of the subject with the therapeutic cell composition. In some embodiments, the pharmacokinetic response is a measure of maximum CAR T cell concentration in the subject following treatment of the subject with the therapeutic cell composition. In some embodiments, the pharmacokinetic response is a measure of a timepoint at which CAR T cell concentration is maximal in the subject following treatment of the subject with the therapeutic cell composition. In some embodiments, the pharmacokinetic response is a measure of exposure of the subject to CAR T cells of the therapeutic cell composition following treatment of the subject with the therapeutic cell composition. [0033] In some embodiments, the sample comprises a whole blood sample, a buffy coat sample, a peripheral blood mononuclear cell (PBMC) sample, an unfractionated T cell sample, a lymphocyte sample, a white blood cell sample, an apheresis product, or a leukapheresis product. In some embodiments, the sample is an apheresis product or leukapheresis product. In some embodiments, the apheresis product or leukapheresis product has been previously cryopreserved. In some embodiments, the T cells comprise primary cells obtained from the subject. In some embodiments, the T cells comprise CD3+, CD4+, and/or CD8+. [0034] In some embodiments, the input composition comprises CD4+, CD8+, or CD4+ and CD8+ T cells and the therapeutic cell composition comprises CD4+, CD8+, or CD4+ and CD8+ T cells expressing a recombinant receptor and is produced from the input composition, wherein the input composition features comprise input composition features from the CD4+, CD8+, or CD4+ and CD8+ T cell compositions of the input composition, and the therapeutic cell composition features comprise therapeutic cell composition features from the CD4+, CD8+, or CD4+ and CD8+ T cells of the therapeutic composition. [0035] In some embodiments, the input composition comprises separate compositions of CD4+ and CD8+ T cells and the therapeutic cell composition comprises separate compositions of CD4+ and CD8+ T cells expressing a recombinant receptor, and is produced from the respective CD4+ or CD8+ T cell composition of the input composition, wherein the input composition features comprise input composition features from the CD4+ and CD8+ T cell compositions of the input composition, and the therapeutic cell composition features comprise therapeutic cell composition features from the CD4+ and CD8+ T cells of each of the separate compositions of the therapeutic composition. [0036] In some embodiments, the input composition comprises separate compositions of CD4+ and CD8+ T cells and the therapeutic cell composition comprises a mixed composition of CD4+ and CD8+ T cells expressing a recombinant receptor, and is produced from the separate CD4+ and CD8+ T cell compositions of the input composition, wherein the input composition features comprise input composition features from the separate CD4+ and CD8+ T cell compositions of the input composition, and the therapeutic cell composition features comprise therapeutic cell composition features from the mixed composition of CD4+ and CD8+ cells of the therapeutic composition. [0037] In some embodiments, the recombinant receptor is a chimeric antigen receptor (CAR). [0038] In some embodiments, the predetermined treatment regimen comprises or is a single treatment comprising administering: a) 25×10.sup.6 CD8+CAR+ T cells and 25×10.sup.6 CD4+CAR+ T cells separately to the subject; b) 50×10.sup.6 CD8+CAR+ T cells and 50×10.sup.6 CD4+CAR+ T cells separately to the subject; or c) 75×10.sup.6 CD8+CAR+ T cells and 75×10.sup.6 CD4+CAR+ T cells separately to the subject. In some embodiments, altering the predetermined treatment regimen, e.g., the altered treatment regimen, comprises or is a single treatment comprising administering: 50×10.sup.6 CD8+CAR+ T cells and 50×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 25×10.sup.6 CD8+CAR+ T cells and 25×10.sup.6 CD4+CAR+ T cells separately to the subject; 75×10.sup.6 CD8+CAR+ T cells and 75×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 50×10.sup.6 CD8+CAR+ T cells and 50×10.sup.6 CD4+CAR+ T cells separately to the subject; or 75×10.sup.6 CD8+CAR+ T cells and 75×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 25×10.sup.6 CD8+CAR+ T cells and 25×10.sup.6 CD4+CAR+ T cells separately to the subject. In some embodiments, altering the predetermined treatment regimen, e.g., the altered treatment regimen, comprises or is a single treatment comprising administering: 25×10.sup.6 CD8+CAR+ T cells and 25×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 50×10.sup.6 CD8+CAR+ T cells and 50×10.sup.6 CD4+CAR+ T cells separately to the subject; 50×10.sup.6 CD8+CAR+ T cells and 50×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 75×10.sup.6 CD8+CAR+ T cells and 75×10.sup.6 CD4+CAR+ T cells separately to the subject; or 25×10.sup.6 CD8+CAR+ T cells and 25×10.sup.6 CD4+CAR+ T cells separately to the subject when the predetermined treatment regimen comprises or is a single treatment comprising administering 75×10.sup.6 CD8+CAR+ T cells and 75×10.sup.6 CD4+CAR+ T cells separately to the subject. In some embodiments, wherein altering the predetermined treatment regimen, e.g., the altered treatment regimen, comprises administering the therapeutic cell composition in combination with a second therapeutic agent. |
return to message board, top of board |