Pre-determined genomic adjusted radiation dose (GARD) values for oropharyngeal cancer

Abstract

Disclosed herein are systems and methods for personalized treatment of individual patient tumor. In one embodiment, a computer software configured to integrate with a radiation therapy treatment planning system can be configured to obtain expression levels of signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors, assign a radiation sensitivity index (RSI) of the subject's tumor based at least in part on the expression levels of the signature genes in the tumor, calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer, and provide the recommended RxRSI as a radiation therapy dose for a radiation plan.

Description

Attorney Docket No. 146974.000006PRE-DETERMINED GENOMIC ADJUSTED RADIATION DOSE (GARD) VALUES FOR OROPHARYNGEAL CANCERCROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of, and priority under 35 U. S. C. § 119(e) to U. S. Provisional Patent Application No. 63 / 760,495 filed February 19, 2025, the entire content of which is incorporated by reference herein.

[0002] This application is related to International Patent Application No. PCT / US24 / 43451, filed August 22, 2024, entitled “Predetermined Genomic Adjusted Radiation Dose (GARD) Values For Nasopharyngeal And Oropharyngeal Cancers.” This application is also related to U. S. Patent Application No. 18 / 538,477 (now U. S. Patent No. 12,226,655), filed December 13, 2023, which is a continuation of U. S. Patent Application No. 18 / 148,502 (now U. S. Patent No.11,865,365), filed December 30, 2022, which is a continuation of U. S. Patent Application No.16 / 658,961, filed October 21, 2019 (now U. S. Patent No. 11,547,871), which claims priority to U. S. Provisional Patent Application No. 62 / 747,861, filed on October 19, 2018, entitled “Systems and Methods for Personalized Radiation Therapy.” Each of the applications referenced in this paragraph are incorporated by reference in their entireties.SUMMARY

[0003] This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0004] Disclosed herein are systems and methods for personalized treatment of individual patient tumors.

[0005] According to an aspect of the present disclosure, a computer-implemented method for personalizing radiation therapy based on clinical treatment priority is provided. The method includes obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method further includes obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor. The method includes determining a clinical treatmentAttorney Docket No.: 146974.000006 PATENTpriority selected from a first clinical treatment priority and a second clinical treatment priority. The method includes selecting a pre -determined genomic adjusted radiation dose (GARD) value based on the determined clinical treatment priority, wherein a first GARD value is associated with the first clinical treatment priority and a second GARD value, different from the first GARD value, is associated with the second clinical treatment priority. The method includes calculating a radiation dosage (RxRSI) for the subject based at least in part on the RSI and the selected pre-determined GARD value.

[0006] According to another aspect of the present disclosure, a method of treating a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of the same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes selecting a clinical treatment priority. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being selected as a function of the selected clinical treatment priority. The method includes providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the clinical treatment priority. The method includes storing treatment outcome data for the subject based on a treatment outcome of the personalized radiation therapy treatment plan for the subject to determine one or more of: an on-target efficacy for a reference population of subjects, an off- target toxicity effect for the reference population of subjects.

[0007] According to another aspect of the present disclosure, a method of identifying a subject with oropharyngeal cancer for radiation therapy dose de-escalation is provided. The method includes obtaining expression levels of one or more signature genes from a tumor sample of the subject. The method includes determining a radiation sensitivity index (RSI) of the subject's tumor sample based at least in part on the expression levels. The method includes calculating a genomic adjusted radiation dose (GARD) value for the subject based at least in part on the RSI and a proposed radiation dose. The method includes comparing the calculated GARD value to a pre-determined GARD threshold. The method includes identifying the subject as a candidate for dose de-escalation when the calculated GARD value meets or exceeds the pre-determined GARD threshold. The method includes generating a radiation therapy treatment plan for the subject based at least in part on the identification, wherein theAttorney Docket No.: 146974.000006 PATENTradiation therapy treatment plan specifies a de-escalated radiation dose for subjects identified as candidates for dose de-escalation.

[0008] According to another aspect of the present disclosure, a computer software configured to integrate with a radiation therapy treatment planning system is provided. The computer software is configured to obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The computer software is configured to assign a radiation sensitivity index (RSI) of the subject’s tumor based at least in part on the expression levels of the one or more signature genes in the tumor. The computer software is configured to calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer. The computer software is configured to provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0009] According to another aspect of the present disclosure, a computer-implemented method for minimizing risk of radiation therapy is provided. The method includes obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on tire RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer. The method includes providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0010] According to another aspect of the present disclosure, a method of calculating a personalized radiation therapy dosage for a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumorAttorney Docket No.: 146974.000006 PATENTsample. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

[0011] According to another aspect of the present disclosure, a method of treating a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

[0012] According to another aspect of the present disclosure, a computer software configured to integrate with a radiation therapy treatment planning system is provided. The computer software is configured to obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The computer software is configured to assign a radiation sensitivity index (RSI) of tire subject's tumor based at least in part on the expression levels of the one or more signature genes in the tumor. The computer software is configured to calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer. The computer software is configured to provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0013] According to another aspect of the present disclosure, a computer-implemented method for minimizing risk of radiation therapy is provided. The method includes obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor. TheAttorney Docket No.: 146974.000006 PATENTmethod includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer. The method includes providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0014] According to another aspect of the present disclosure, a method of calculating a personalized radiation therapy dosage for a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre -determined GARD value being about 32 for oropharyngeal cancer.

[0015] According to another aspect of the present disclosure, a method of treating a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer.

[0016] According to another aspect of the present disclosure, a computer software configured to integrate with a radiation therapy treatment planning system is provided. The computer software is configured to obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The computer s il v are is configured to assign a radiation sensitivity index (RSI) of the subject’s tumor based at least in part on the expression levels of the one or more signature genes in the tumor. The computer software is configured to calculate a recommended personalized radiation dosage (RxRSI) for the subjectAttorney Docket No.: 146974.000006 PATENTbased at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer. The computer software is configured to provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0017] According to another aspect of the present disclosure, a computer-implemented method for minimizing risk of radiation therapy is provided. The method includes obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of tire one or more signature genes in tire tumor. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer. The method includes providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0018] According to another aspect of the present disclosure, a method of calculating a personalized radiation therapy dosage for a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

[0019] According to another aspect of tire present disclosure, a method of treating a subject is provided. The method includes determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type. The method includes applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample. The method includes calculating a personalized radiationAttorney Docket No.: 146974.000006 PATENTdosage (RxRSI) for the subject based at least in part on die RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

[0020] According to another aspect of the present disclosure, a system for providing a radiation therapy treatment plan is provided. The system includes one or more processors and a memory storing instructions that, when executed by the one or more processors, cause the system to obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor associated with oropharyngeal cancer or selected from a database of previously measured gene-expression profiles for a reference population of tumors associated with oropharyngeal cancer. The instructions cause the system to assign a radiation sensitivity index (RSI) to the tumor associated with oropharyngeal cancer based at least in part on the expression levels of the one or more signature genes in the tumor. The instructions cause the system to calculate a recommended personalized radiation dosage (RxRSI) for the subject with the tumor based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value. The instructions cause the system to assign a minimal clinical target GARD value of 32. The instructions cause the system to assign a recommended effective target GARD value of 41.9.

[0021] The foregoing general description of the illustrative embodiments and the following detailed description thereof are merely exemplary aspects of the teachings of this disclosure and are not restrictive.BRIEF DESCRIPTION OF THE FIGURES

[0022] The file of this patent contains at least one drawing / photograph executed in color. Copies of tliis patent with color drawing(s) / photograph(s) will be provided by the Office upon request and payment of the necessary fee.

[0023] FIG. 1A shows GARD plotted against EQD2 (Gy), according to an exemplary embodiment of the present disclosure.

[0024] FIG. IB shows GARD plotted against EQD2 (Gy), according to an exemplary embodiment of the present disclosure.

[0025] FIG. 2A shows log relative hazard plotted against GARD, according to an exemplary embodiment of the present disclosure.

[0026] FIG. 2B shows log relative hazard plotted against GARD, according to an exemplary embodiment of the present disclosure.Attorney Docket No.: 146974.000006 PATENT

[0027] FIG. 3A shows a plot of survival probability over time, according to an exemplary embodiment of the present disclosure.

[0028] FIG. 3B shows a plot of survival probability over time, according to an exemplary embodiment of the present disclosure.

[0029] FIG. 3C shows a plot of survival probability over time, according to an exemplary embodiment of the present disclosure.

[0030] FIG. 3D shows a plot of survival probability over time, according to an exemplary embodiment of the present disclosure.

[0031] FIG. 4 shows a nomogram incorporating GARD, TNM8, and a 3-cluster prognostic model, according to an exemplary embodiment of the present disclosure.

[0032] FIG. 5 shows a plot of sensitivity versus specificity, according to an exemplary embodiment of the present disclosure.

[0033] FIG. 6A shows a plot of the percentage of patients free of local recurrence over time when a uniform radiation dose of 60 Gy is applied, demonstrating a decrease in recurrence-free survival compared to the standard 70 Gy dose, according to an exemplary embodiment of the present disclosure.

[0034] FIG. 6B shows a plot of percentage of patients free of local recurrence over time when a uniform radiation dose of 60 Gy is applied with selective de-intensification based on GARD values, illustrating that patient-specific dose adjustments can maintain recurrence-free survival comparable to the standard dose, according to an exemplary embodiment of the present disclosure.

[0035] FIG. 7 illustrates a block diagram of an illustrative data processing system, according to an exemplary embodiment of the present disclosure.

[0036] FIG. 8 illustrates patient selection for HPV-positive oropharyngeal squamous cell carcinoma patients, according to an exemplary embodiment of the present disclosure.

[0037] FIG. 9 illustrates a histogram of GARD values, stratified by high and low GARD groups, according to an exemplary embodiment of the present disclosure.

[0038] FIG. 10 depicts a scatterplot showing the relationship between personalized radiation sensitivity index (RxRSI) in Gy and GARD values derived from the associated radiation dose, according to an exemplary embodiment of the present disclosure.

[0039] FIG. 11 shows EQD2 plotted against associated GARD, according to an exemplary embodiment of the present disclosure.

[0040] FIG. 12 A shows log relative hazard plotted against GARD, according to an exemplary embodiment of the present disclosure.Attorney Docket No.: 146974.000006 PATENT

[0041] FIG. 12B shows log relative hazard plotted against GARD, according to an exemplary embodiment of the present disclosure.

[0042] FIG. 13 shows a plot of sensitivity versus specificity, according to an exemplary embodiment of the present disclosure.

[0043] FIG. 14 shows box plots illustrating the distribution of GARD values across three patient clusters, according to an exemplary embodiment of the present disclosure.

[0044] FIG. 15 shows comparative histograms of radiation sensitivity index (RSI) and GARD values stratified by AJCC8 stage (I, II, and III), according to an exemplary embodiment of the present disclosure.

[0045] FIG. 16 shows Kaplan-Meier survival curves comparing unselected dose de-escalation strategies based on in silico simulations for HPV-positive oropharyngeal cancer patients, according to an exemplary embodiment of the present disclosure.

[0046] FIG. 17 shows Kaplan-Meier survival curves comparing GARD-selected dose de-escalation strategies based on in silico simulations for HPV-positive oropharyngeal cancer patients, according to an exemplary embodiment of the present disclosure.

[0047] FIG. 18 shows in silico clinical trial designs comparing unselected radiotherapy dose de-escalation to GARD-based stratified de-escalation in HPV-positive oropharyngeal cancer patients, according to an exemplary embodiment of the present disclosure.

[0048] FIG. 19 shows a plot of survival probability over time, according to an exemplary embodiment of the present disclosure.

[0049] FIG. 20 shows a histogram depicting the distribution of radiation doses required to achieve equipoise in terms of clinical outcomes for HPV-positive oropharyngeal cancer patients, according to an exemplary embodiment of the present disclosure.DETAILED DESCRIPTION

[0050] Various aspects now will be described more fully hereinafter. Such aspects may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey its scope to those skilled in the art.

[0051] Where a range of values is provided, it is intended that each intervening value between the upper and lower limit of that range and any other stated or intervening value in that stated range is encompassed within the disclosure. For example, if a range of 1 pm to 8 pm is stated, it is intended that 2 pm, 3 pm, 4 pm, 5 pm, 6 pm, and 7 pm are also explicitly disclosed, as well as the range of values greater than or equal to 1 pm and the range of valuesAttorney Docket No.: 146974.000006 PATENTless than or equal to 8 pm. Further, referring to ranges between certain values, for from a first value to a second value, means the endpoints are to be included. For example, reference to a GARD value from about X to about Y is inclusive of both X and Y, as well as all values between X and Y. The term “about” is further defined herein.

[0052] The singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of embodiments disclosed, the preferred methods, devices, and materials are now described.

[0053] The transitional term “comprising,” which is synonymous with “including,” “containing,” or “characterized by,” is inclusive or open-ended and does not exclude additional, unrecited elements or method steps. By contrast, the transitional phrase “consisting of’ excludes any element, step, or ingredient not specified in the claim. The transitional phrase “consisting essentially of” limits the scope of a claim to the specified materials or steps “and those that do not materially affect the basic and novel characteristic(s)” of the claimed invention. In embodiments or claims where the term comprising is used as the transition phrase, such embodiments can also be envisioned with replacement of the term “comprising” with the terms “consisting of’ or “consisting essentially of.”

[0054] The term “patient” and “subject” are interchangeable and may be taken to mean any- living organism which may be treated with compounds of the present invention. As such, the terms “patient” and “subject” may include, but are not limited to, any non-human mammal, primate or human. In some embodiments, the “patient” or “subject” is a mammal, such as mice, rats, other rodents, rabbits, dogs, cats, swine, cattle, sheep, horses, primates, or humans. In some embodiments, the patient or subject is an adult, child, or infant. In some embodiments, the patient or subject is a human.

[0055] The term “treating” is used herein, for instance, in reference to methods of heating an oropharyngeal disorder or a systemic condition, and generally includes the administration of a compound or composition or a therapy regimen which reduces the frequency of, or delays the onset of, symptoms of a medical condition. This can include reversing, reducing, or arresting the symptoms, clinical signs, and underlying pathology of a condition in a manner to improve or stabilize a subject’s condition.

[0056] As is outlined in greater detail below, when reference is made to “about” a certain GARD value, it is understood that “about” means within a range of + / - 10%.Attorney Docket No.: 146974.000006 PATENT

[0057] Radiation therapy (RT) is the medical use of radiation to treat malignant cells, such as cancer cells. This radiation can have an electromagnetic form, such as a high-energy photon, or a particulate form, such as an electron, proton, neutron, or alpha particle. By far, the most common form of radiation used in practice today is high-energy photons. Photon absorption in human tissue is determined by the energy of the radiation, as well as the atomic structure of the tissue in question. The basic unit of energy used in radiation oncology is the electron volt (eV); 103 eV=1 keV, 106 eV=1 MeV. At therapeutic energies, the three major interactions between photons and tissue are the photoelectric effect, Compton effect, and pair production.

[0058] Due to biological heterogeneity, RT does not uniformly work on all tissue samples, and a uniform “one-size fits all” RT dose for a given cancer-type may not be ideal. Therefore, there remains a need for personalized radiation dose planning methods and systems.

[0059] Disclosed herein are systems and methods for personalized treatment of individual patient tumor. In one embodiment, a computer software configured to integrate with a radiation therapy treatment planning system can be configured to assign a radiation sensitivity index (RSI) of a subject’s tumor based at least in part on expression levels of one or more signature genes in the tumor, calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer, and provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan. In an alternative embodiment, the pre-determined GARD value can be about 32 for oropharyngeal cancer. In an alternative embodiment, the pre-determined GARD value can be from about 32 to about 41.9 for oropharyngeal cancer.

[0060] The computer software can be further configured to calculate the recommended RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

[0061] The computer software can be further configured to calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

[0062] The computer software can be further configured to receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the recommended RxRSI, calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans, penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the radiation treatment plan, and provide, to the radiation therapy treatment planningAttorney Docket No.: 146974.000006 PATENTsystem, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

[0063] The computer software can be further configured to calculate the recommended RxRSI based in part on a predefined standard of care dose range.

[0064] The computer software can be further configured to calculate a proposed RxRSI based for the subject based at least in part on the pre-determined GARD value and the RSI, compare the proposed RxRSI to a predefined standard of care dose range, assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range, and recommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

[0065] The computer software can be further configured to apply a linear regression model to the expression levels of the one or more signature genes in the tumor and assign the RSI based at least in part on the linear regression model.

[0066] The pre-determined GARD value can be based at least in part on a plurality of GARD values for subjects in a cohort.

[0067] In an additional embodiment, a computer-implemented method for minimizing the risk of radiation therapy can include obtaining an RSI of a subject's tumor from expression levels of one or more signature genes in the tumor, calculating an RxRSI for the subject based at least in part on the RSI and a pre-determined GARD value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer, and providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0068] The computer-implemented method can further include calculating normal tissue toxicity of the personalized radiation dosage and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity. In an embodiment, the normal tissue toxicity is based at least in part on risks to a plurality of tissue sites, such as for example and not limitation, esophagus, lung, heart, small bowel, brain, rectum, bladder, spinal cord, kidney, or skin.

[0069] The computer-implemented method can provide a plurality of radiation treatment plans, which can be penalized based on tissue toxicity to arrive at an optimal radiation treatment plan that has the least tissue toxicity.

[0070] In an additional embodiment, a method of developing a personalized radiation treatment plan can include assigning an RSI of a subject’s tumor based at least in part on expression levels of one or more signature genes in the tumor, calculating an RxRSI for theAttorney Docket No.: 146974.000006 PATENTsubject based at least in part on a pre-determined GARD value and the RSI, and providing the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0071] The method can further include calculating the recommended RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

[0072] The method can further include calculating the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

[0073] The method can further include calculating a respective normal tissue toxicity for each radiation plan of a plurality of radiation plans each utilizing the recommended RxRSI, penalizing each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the radiation treatment plan, and providing at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

[0074] The method can further include determining whether the recommended RxRSI is within a predefined standard of care dose range.

[0075] The method can further include calculating a proposed RxRSI based for the subject based at least in part on the pre-determined GARD value and the RSI, comparing the proposed RxRSI to a predefined standard of care dose range, assigning the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range, and recommending consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

[0076] The method can further include applying a linear regression model to the expression levels of the one or more signature genes in the tumor and assigning the RSI based at least in part on the linear regression model.

[0077] The pre-determined GARD value is based at least in part on a plurality of GARD values for subjects in a cohort.

[0078] In a further embodiment, a method of calculating a personalized radiation therapy dosage for a subject can include determining expression levels of one or more signature genes from a subjects tumor sample, applying a linear regression model to the expression levels and assigning an RSI to the subject's tumor sample, and calculating an RxRSI for the subject based at least in part on the RSI and a pre-determined GARD value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

[0079] In a further embodiment, a method of treating a subject can include determining expression levels of one or more signature genes from a subject’s tumor sample, applying a linear regression model to the expression levels and assigning an RSI to the subject's tumor sample, calculating an RxRSI for the subject based at least in part on the RSI and a pre-Attorney Docket No.: 146974.000006 PATENTdetermined GARD value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer, and administering the calculated RxRSI to the subject as a treatment for oropharyngeal cancer.

[0080] In some embodiments, any method known in the art may be used for obtaining a tumor sample from a subject. The tumor sample may comprise at least one living cell (preferably a plurality of cells), e.g., a cell from a tumor (e.g., from a biopsy), a normal cell, or a cultured cell. Commonly used methods to obtain tumor cells include surgical (the use of tissue taken from the tumor after removal of all or part of the tumor) and needle biopsies. The samples should be treated in any way that preserves intact the expression levels of the living cells as much as possible, e.g., flash freezing or chemical fixation, e.g., formalin fixation. Any method known in the art can be used to extract material, e.g., protein or nucleic acid (e.g., mRNA) from the sample. For example, mechanical or enzymatic cell disruption can be used, followed by a solid phase method (e.g., using a column) or phenol -chloroform extraction, e.g., guanidinium thiocyanate-phenol-chloroform extraction of the RNA. A number of kits are commercially available for use in isolating mRNA. Purification can also be used if desired.

[0081] In some embodiments, the tumor is a cancer tumor selected from oropharyngeal cancer. In some embodiments, the tumor is a cancer tumor selected from colorectal cancer, breast cancer, ovarian cancer, pancreatic cancer, head and neck cancer, bladder cancer, liver cancer, renal cancer, melanoma, gastrointestinal cancer, prostate cancer, small cell lung cancer, non-small cell lung cancer, sarcoma, glioblastoma, T-cell lymphoma, B-cell lymphoma, endometrial cancer, and cervical cancer.

[0082] In some embodiments, any method known in the art may be used to determine the expression levels in a tumor sample. Gene expression levels can be determined in many different ways including the quantification of fluorescence of hybridized mRNA on glass slides, Northern blot analysis, real-time reverse transcription PCR (RT-PCR), microarray or other measures of gene expression abundance.

[0083] In some embodiments, the methods include determining expression levels of signature genes in one or more cells of a tumor. In some embodiments, the methods include determining the expression levels of a plurality of signature genes, e.g., two, three, four, five, six, seven, eight, nine, or all ten signature genes, as follows: androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STATl); protein kinase C, beta (PKCb); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase I (HDAC1); and interferon regulatory factor 1 (IRFl ).Attorney Docket No.: 146974.000006 PATENT

[0084] In some embodiments, the methods include determining expression levels of signature genes in one or more cells of a tumor, and determining an RSI of the tumor based on the expression levels of the signature genes. To determine the RSI, the methods described herein may use a rank-based linear algorithm.

[0085] In some embodiments, determining a radiation sensitivity index of a tumor comprises applying a linear regression model to the gene expression levels, e.g., a rank-based linear regression model. In some embodiments, the expression levels of the plurality of signature genes are weighted. A linear regression model useful in the methods described herein includes gene expression levels and coefficients, or weights, for combining expression levels. The coefficients can be calculated using a least-squares fit of the proposed model to a measure of cellular radiation sensitivity. The functional form of the algorithm is given below, where each of the kj coefficients will be determined by fitting expression levels to a particular RSI measure:RSI=k1*AR+k2*c-jun+k3*STAT1+k4*PKC+k5*RelA+k6*cAbl+k7*SUMO1+k8*PAK2+k9*HDAC+k10*IRF1

[0086] Further methods and embodiments for determining RSI are described in U. S. Patent Nos. 8,660,801; 9,846,762; and 8,655,598, which are incorporated herein by reference.

[0087] As described herein, RSI provides an indication of whether radiation therapy is likely to be effective in treating the subject's tumor. RSI has a value approximately between 0 and 1. It should be understood that assigning RSI according to the linear regression model of gene expression levels described in U. S. Pat. Nos. 8,660,801; 9,846,762; and 8,655,598, is provided only as an example and that other known techniques for assigning radiation sensitivity can optionally be used with the systems and methods described herein.

[0088] In some embodiments, the method further comprises calculating the GARD for each tumor sample, and is described in U. S. Patent Application No. 15 / 571,617, granted as U. S. Patent No. 10,697,023, which is incorporated herein by reference. Further related methods and embodiments for determining RSI and / or calculating GARD are included in PCT Application No. PCT / US2024 / 043451 (Attorney Docket No. 146974.000003), filed on August 22, 2024, publication of which is forthcoming. GARD is derived using the linear quadratic (LQ) model, the individual RSI and the radiation dose and fractionation schedule for each patient as follows:

[0089] The LQ model in its simplest form is represented by:S=e-nd(α+βd)where n is the number of fractions of radiation, d is the dose per fraction, and a and represent the linear and quadratic radiosensitivity parameters, respectively.Attorney Docket No.: 146974.000006 PATENT

[0090] Since RSI is a molecular estimate of SF2 in cell lines (survival fraction at 2 Gy), a patient-specific a is derived by substituting RSI for Survival (S) in the equation above, where dose (d) is 2 Gy, n=l and p is a constant (0.05 / Gy2). GARD is calculated using the classic equation for biologic effect shown by equation E=nd(a+(3d), the patient-specific a and the radiation dose and fractionation received by each patient. Additionally, the GARD value can be predictive of tumor recurrence in the subject after treatment.

[0091] In some embodiments, the method comprises calculating an RxRSI for each individual tumor or subject based on a pre-determined GARD value. The RxRSI is the dose required to achieve a pre-determined GARD value. The RxRSI is calculated using the formula below:RxRSI = GARD target value / (α + βd),where a is calculated based on the patient’s RSI as described above and 3 is a constant (0.05 / Gy2).

[0092] In some embodiments, a pre-determined GARD value may be calculated based on improved outcome in a particular cancer type. In other embodiments, a pre-determined GARD value may be calculated based on empiric values for a cancer type.

[0093] The pre-determined GARD value may vary depending on the cancer type. For example, the pre-determined GARD value for a subject suffering from oropharyngeal cancer may be between about 41.9 or about 32 for oropharyngeal cancer. In some embodiments, the pre-determined GARD value for other cancers may be more or less than about 32, such as any number between 2 and 150.

[0094] HPV-positive oropharyngeal cancer patients, for example, generally have excellent prognosis, which led to the clinical hypothesis that these patients could safely receive reduced radiation doses based on favorable clinical factors. This hypothesis was tested in a prospective clinical trial, which evaluated uniform radiation dose de-escalation from 70 Gy to 60 Gy in HPV-positive oropharyngeal cancer patients. The trial failed to meet its non-inferiority endpoint, demonstrating that uniform dose de-escalation based on clinical factors alone results in inferior clinical outcomes. Analysis using GARD revealed that this failure can be explained by the underlying genomic heterogeneity in radiosensitivity among HPV-positive patients: while the majority of patients can safely receive lower doses, a subset of patients with radioresistant tumors require standard or higher doses to maintain tumor control. GARD provides a mechanism to identify these patient subsets and enable safe, selective dose de-escalation.Attorney Docket No.: 146974.000006 PATENT

[0095] Through analysis of HPV-positive oropharyngeal cancer patient cohorts, it was discovered that a GARD value of about 32 achieves clinical equipoise with current standard of care outcomes. Patients who achieve a GARD of at least 32 demonstrate overall survival rates equivalent to the unselected population treated with standard 70 Gy dosing. This equipoise threshold is distinct from the optimal outcome-maximizing threshold of about 41.9 or 42, which is associated with the highest probability of favorable overall survival. The identification of these two thresholds enables clinicians to select a GARD target based on the desired clinical treatment priority: selecting about 41.9 or 42 to maximize individual cure probability, or selecting about 32 to maintain population-level outcomes while enabling dose reduction.

[0096] Using the GARD 32 equipoise threshold, approximately 77.7% of oropharyngeal cancer patients become candidates for dose de-escalation, with an average reduction of approximately 5 fractions (equi valent to one week of radiotherapy) per patient. The remaining 22.3% of patients are identified as potentially requiring standard or escalated doses. This genomic -based stratification achieves what uniform dose de-escalation trials could not: equivalent clinical outcomes at reduced average dose, by accounting for the biological heterogeneity that uniform dosing strategies fail to address.

[0097] In some embodiments, the pre-determined GARD value may be selected from a range of values based on the desired clinical outcome. A first GARD value may be associated with a first clinical treatment priority of maximizing therapeutic effect, while a second GARD value, lower than the first GARD value, may be associated with a second clinical treatment priority of maintaining acceptable therapeutic outcomes while reducing radiation dose and associated toxicity. The specific numeric values of the first and second GARD values may vary depending on the cancer type, patient population characteristics, and clinical goals. In some embodiments, the first GARD value and the second GARD value may be determined empirically from clinical outcome data for a reference population.

[0098] In some embodiments, the RSI for use in calculating an RxRSI can be obtained from a reference database that houses gene-expression data or previously computed RSI values for a population of subjects who share relevant tumor characteristics (e.g., the same cancer type, stage, or biomarker profile). This approach may be referred to as “Populational GARD.”

[0099] In some embodiments, the system or method may access a database containing expression profiles or RSI values computed from historical clinical trials, population-wide repositories, or other aggregated sources. For instance, if a new subject’s tumor gene¬ expression data is not immediately available, or the subject’s tumor sample is of insufficient quality for gene-expression analysis, the software may: identify a reference subset in theAttorney Docket No.: 146974.000006 PATENTdatabase whose tumor type is similar or identical to the subject’s tumor (e.g., same histological subtype, same HPV status, oropharyngeal cancer, etc.); retrieve one or more representative RSI values from that database, for example a median RSI, a mean RSI, or an RSI distribution from the matched reference subset; and / or select or calculate a single RSI for the subject’s tumor from that distribution (e.g., average, best-fit, percentile rank, etc.).

[0100] In some embodiments, the system may combine partial subject-specific geneexpression data (even if incomplete) with population reference data to derive a more refined RSI estimate. For example, when only a panel of gene-expression data is available from the subject’s tumor, the system can supplement that partial data with population-derived distributions for the missing genes. The resulting composite RSI is still used in the same manner to calculate a recommended dosage (RxRSI).

[0101] In these “Populational GARD” embodiments, the software or method may proceed to obtain the population-based RSI (or partial gene-expression data from the population) and treat that as if it were the subject’s own RSI; calculate a recommended radiation dosage (RxRSI) based on that RSI using the same formulas described herein for GARD; and / or provide the resulting RxRSI to the radiation therapy treatment planning system,

[0102] As used herein, the term “subject” may refer to an individual patient, in which case “assigning RSI” is performed on expression levels measured directly from that patient’s tumor sample, or a representative or aggregated data set for a population of subjects, such that “assigning RSI” can be performed on gene-expression data (or RSI values) previously- generated and stored in a database.

[0103] References to “the tumor” or “the subject’s tumor” may encompass both directly measured tumor data from an individual patent and previously gathered data or RSI values from population databases that are used as a proxy or reference for the new patient’s tumor.

[0104] The empirical radiation dose for a solid epithelial tumor ranges from 60 to 80 Gy, while lymphomas are treated with 20 to 40 Gy. For example, lung cancers are treated between 60 and 74 Gy, prostate cancers are generally treated between 37.25 to 80 Gy, esophageal cancers are treated between 44 to 70 Gy, oropharyngeal cancers are treated between 60 to 70 Gy, and nasopharyngeal cancers are treated between 66 to 70 Gy. It is possible that the empirical dose that the patients receive is lower or higher than what they need. A personalized radiation dose would be ideal to achieve an improved outcome.

[0105] In some embodiments, the personalized radiation dose that is calculated may be 5% less than the empirical dosing value, may be 10% less than the empirical dosing value, may be 15% less than the empirical dosing value, may be 20% less than the empirical dosing value,Attorney Docket No.: 146974.000006 PATENTmay be 25% less than tlie empirical dosing value, may be 30% less than the empirical dosing value, may be 35% less than the empirical dosing value, may be 40% less than the empirical dosing value, may be 50% less than the empirical dosing value, or may be 60% less than the empirical dosing value.

[0106] In some embodiments, the personalized radiation dose that is calculated may be 5% more than the empirical dosing value, may be 10% more than the empirical dosing value, may be 15% more than the empirical dosing value, may be 20% more than the empirical dosing value, may be 25% more than the empirical dosing value, may be 30% more than the empirical dosing value, may be 35% more than the empirical dosing value, may be 40% more than the empirical dosing value, may be 50% more than the empirical dosing value, or may be 60% more than tlie empirical dosing value.

[0107] For clarity, these percentage variations apply to the empirical radiation dosing values and do not modify the pre-determined genomic-adjusted radiation dose (GARD) values, which remain within tlie defined range of + / - 10% as described above. The calculated GARD value remains a guiding factor in determining the personalized radiation dose within clinically appropriate limits.

[0108] In some embodiments, radiation is administered in at least about 1 Gray (Gy) fraction at least once every other day to a treatment volume. In some embodiments, radiation is administered in at least about 2 Gy fractions at least once per day to a treatment volume. In some embodiments, radiation is administered in at least about 2 Gy fractions at least once per day to a treatment volume for five consecutive days per week. In another embodiment, radiation is administered in 3 Gy fractions every other day, three times per week to a treatment volume. In yet another embodiment, a total of at least about 20 Gy, about 30 Gy, about 40 Gy, about 50 Gy, about 60 Gy, about 70 Gy, about 80 Gy, about 90 Gy, or about 100 Gy of radiation is administered to a subject in need thereof.

[0109] The methods disclosed herein may be practiced in an adjuvant setting. “Adjuvant seting” refers to a clinical setting in which an individual has a history of a proliferative disease, particularly cancer, and generally (but not necessarily) has been treated with therapy, which includes, but is not limited to, surgery and / or chemotherapy. However, because of a history of the proliferative disease, these individuals are considered at risk of developing that disease or may harbor detectable and / or microscopic disease. Treatment or administration in the “adjuvant setting” refers to a subsequent mode of treatment.

[0110] The methods provided herein may also be practiced in a “neoadjuvant setting,” that is, the method may be carried out before the primary / definitive therapy. In some respects, theAttorney Docket No.: 146974.000006 PATENTindividual has previously been treated. In other aspects, the individual has not previously been treated. In some respects, the treatment is a first line therapy.

[0111] In some embodiments, any of the methods of treatment of RT described herein can be administered in combination with one or more additional therapies to the individual, such as surgery and / or chemotherapy. In some embodiments, various classes of chemotherapeutic agents can be administered in combination with RT. Non-limiting examples include: alkylating agents (e.g. cisplatin, carboplatin, or oxaliplatin), antimetabolites (e.g., azatliioprine or mercaptopurine), anthracyclines, plant alkaloids (including, e.g. vinca alkaloids (such as, vincristine, vinblastine, vinorelbine, or vindesine) and taxanes (such as, paclitaxel, taxol, or docetaxel), topoisomerase inhibitors (e.g., camptothecins, irinotecan, topotecan, amsacrine, etoposide, etoposide phosphate, or teniposide), podophyllotoxin (and derivatives thereof, such as etoposide and teniposide), and other antineoplastics (e.g., dactinomycin, doxorubicin, epirubicin, bleomycin, mechlorethamine, cyclophosphamide, chlorambucil, or ifosfamide).

[0112] In some embodiments, an RT treatment disclosed herein may be combined with other targeted therapies, such as immunoconjugates or antibodies coupled to cytotoxic agents. Non¬ limiting cytotoxic agents that can be coupled to an antibody include a chemotherapeutic agent, a growth inhibitory agent, a toxin (e.g., an enzymatically active toxin of bacterial, fungal, plant, or animal origin, or fragments thereof), or a radioactive isotope (e.g., a radioconjugate).

[0113] Also disclosed herein are systems and methods for developing a personalized RT treatment plan for a subject having a tumor. In some embodiments, the system can include one or more processors and a memory operably coupled to the one or more processors. The memory¬ can include computer-executable instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to determine an RSI of the tumor from expression levels of one or more signature genes in the tumor, determine a GARD value based on RSI, radiation dose and fractionation schedule of the patient, calculate an RxRSI for the subject based on a pre-determined GARD value, calculate the normal tissue toxicity of the personalized radiation dosage, and provide the personalized radiation therapy treatment plan for the subject.

[0114] In some embodiments, the system can include a one or more processors and a memory operably coupled to the one or more processors, the memory having computer-executable instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to provide a personalized radiation therapy treatment plan for the subject. The personalized radiation therapy treatment plan for the subject can be based on one or more of the following inputs: an RSI of the tumor from expression levels of one or more signatureAttorney Docket No.: 146974.000006 PATENTgenes in the tumor, a GARD value based on RSI, radiation dose and fractionation schedule of the patient, an RxRSI for the subject based on a pre-determined GARD value, and the normal tissue toxicity of the personalized radiation dosage.

[0115] In some embodiments, the system can include one or more processors and a memory operably coupled to the one or more processors, the memory having computer-executable instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to calculate an RxRSI for the subject based on a pre-determined GARD value, calculate the normal tissue toxicity of the personalized radiation dosage, and provide the personalized radiation therapy treatment plan for the subject.

[0116] In some embodiments, the system can include one or more processors and a memory operably coupled to the one or more processors, the memory having computer-executable instructions stored thereon that, when executed by the one or more processors, cause the one or more processors to provide the personalized radiation therapy treatment plan for the subject based on RxRSI for the subject and the normal tissue toxicity of the personalized radiation dosage.

[0117] In some embodiments, the method includes integrating the prescribed RT dosage into a commercially available radiation treatment planning system that generates a personalized treatment plan based on the patient’s RSI, GARD and RxRSI values. The methods disclosed herein can be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, or in combinations of them.

[0118] Accordingly, as discussed herein, various embodiments may include non-transitory computer-readable media for analyzing health information. In particular, some embodiments may have a health / diagnosis analysis system configured to analyze, examine, search, investigate, consider, evaluate, and / or otherwise process health information and to generate various medical assessments based on the health information. Non-limiting examples of medical assessments include medical diagnoses, medical orders, and / or risk assessments. Health information, as used herein, may include any type of information associated with the health or physical characteristics of a patient, including, but not limited to, name, address, age, gender, demographic information, weight, height, medications, surgeries and other medical procedures (e.g., diagnostic tests, diagnostic imaging tests, or the like), occupation(s), past and current medical conditions, family history, patient description of health condition, healthcare professional description of health condition, and / or symptoms.

[0119] In some embodiments, the analysis process may involve accessing health information associated with a patient and providing a medical assessment based on various analyses of theAttorney Docket No.: 146974.000006 PATENThealth information. In some embodiments, the health / diagnosis analysis system may receive input from a healthcare provider concerning the accuracy, completeness, correctness, or other measure of a medical assessment for use in determining future medical assessments.

[0120] In some embodiments, the information, or data, acquired by the system may generally include all information collected or generated prior to the medical procedure. Thus, for example, information about the patient may be acquired from a patient intake form or electronic medical record (EMR). Examples of patient information that may be collected include, without limitation, patient demographics, diagnoses, medical histories, progress notes, vital signs, medical history information, allergies, and lab results. The data may also include images related to the patient’s area of interest. It should be understood that the images may be captured using any known or future medical imaging device, for example, Magnetic Resonance Imaging (MRI), Computed Tomography (CT), X-ray, ultrasound, or any other modality known in the art. The data may also comprise quality of life data captured from the patient. For example, in one embodiment, a patient may use a software application C‘app”) to answer one or more questionnaires regarding their current quality of life. In a further embodiment, the health information may include demographic, anthropometric, cultural, or other specific traits about a patient that can coincide with activity levels and specific patient activities to customize the surgical plan to the patient. For example, certain cultures or demographics may be more likely to perform a repetiti ve physical task or be exposed to a particular set of environmental factors.

[0121] In a further embodiment, the computer system may refine or improve the diagnosis by adjusting weighted factors and / or modifying one or more determination factors based on outcome data. For example, an embodiment may utilize a closed loop algorithm to perform statistical and machine learning modeling. In certain implementations, the outcome data may include overall survival information, progression-free survival information, response rate to a specific drug, and / or other similar outcome data.

[0122] For example, a procedure for refining weights can involve testing a variety of statistical and machine learning modeling techniques and selecting the one that performs best. For a given set of medical procedures, multiple models may be trained to predict the outcomes. The best model can be selected, or a combination and / or averaging of the best models may be newly generated. In certain implementations, rules can be in place to determine what alterations are made to the system.

[0123] Accordingly, the algorithm / system as described herein may include machine learning and / or other similar statistical-based modeling techniques. For example, the algorithm used may depend on an expected outcome. A processing device can be configured to use a firstAttorney Docket No.: 146974.000006 PATENTprocess or algorithm to calculate refinements to a derived diagnosis based upon a first set of outcome data while also using a second or different algorithm to calculate refinements. Different methods and algorithms may be used to calculate the refined weights in concert or substantially simultaneously. The output of each of the different methods and algorithms can then be compared / further analyzed to determine which output is highest rated, or the output of each method and algorithm can be combined into a combinational metric.

[0124] In some embodiments, the RxRSI may be calculated using a computer program product. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing one or more processors to carry out aspects of the present i nvention.

[0125] The computer readable storage medium can be a non-transitory tangible device that can retain and store instructions for use by an instruction execution device (e.g., one or more processors). The computer readable storage medium may be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a head disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), static random access memory (SRAM), a compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-card(s) or raised structures in a groove having instructions recorded thereon, and / or any suitable combination of the foregoing.

[0126] A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.

[0127] Computer readable program instructions described herein may be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network (LAN), a wide area network (WAN), and / or a wireless network. The network may comprise conductive transmission cables (e.g., copper cables), optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readableAttorney Docket No.: 146974.000006 PATENTprogram instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0128] The one or more processors can process instructions for execution within the computing device, including instructions stored in the memory. The one or more processors can also include separate analog and digital processors. The one or more processors can provide, for example, coordination of the other components of the device, such as the user interface, applications, and wireless communication.

[0129] The one or more processors may communicate with a user through a control interface and / or a display interface coupled to a display. The display can be, for example, a TFT LCD display, an OLED display, or other appropriate display technology. The display interface can comprise appropriate circuitry for driving the display to present graphical and other information to a user. The control interface can receive commands from a user and convert them for submission to the one or more processors. In addition, an external interface can be in communication with processor, so as to enable near area communication of device with other devices.

[0130] In some embodiments, the system includes computer software that integrates information for each individual patient including imaging, genomic and clinical data (i.e., clinical prescription). The system generates a conventional standard of care (SoC) treatment plan as well as a personalized treatment plan that incorporates the individual patient RSI, GARD, RxRSI, and normal tissue toxicity. The physician can then evaluate both plans and choose which one to use for the patient based on standard dose-volume histogram (DVH) metrics of normal tissue and tumor coverage.

[0131] In some embodiments, a computer-implemented method for minimizing the risk of radiation therapy is provided. The method can include obtaining a radiation sensitivity index (RSI) of a subject’s tumor from expression levels of one or more signature genes in the tumor, determining a GARD value based on RSI, radiation dose and fractionation schedule of the subject, calculating an RxRSI for the subject based on a pre-determined GARD value, calculating normal tissue toxicity of the personalized radiation dosage, and providing a personalized radiation therapy treatment plan for the subject.

[0132] FIG. 7 illustrates a block diagram of an illustrative data processing system 700 in which aspects of the illustrative embodiments are implemented. The data processing system 700 is an example of a computer, such as a server or client, in which computer usable code or instructions implementing the process for illustrative embodiments of the present invention are located. In some embodiments, the data processing system 700 may be a server computingAttorney Docket No.: 146974.000006 PATENTdevice. For example, data processing system 700 can be implemented in a server or another similar computing device operably connected to a surgical system. The data processing system 700 can be configured to, for example, transmit and receive information related to a patient and / or a related surgical plan with the surgical system.

[0133] In the depicted example, data processing system 700 can employ a hub architecture including a north bridge and memory controller hub (NB / MCH) 701 and south bridge and input / output (I / O) controller hub (SB / ICH) 702. Processing unit 703, main memory 704, and graphics processor 705 can be connected to the NB / MCH 701. Graphics processor 705 can be connected to the NB / MCH 701 through, for example, an accelerated graphics port (AGP).

[0134] In the depicted example, a network adapter 706 connects to the SB / ICH 702. An audio adapter 707, keyboard and mouse adapter 708, modem 709, read only memory (ROM) 710, hard disk drive (HDD) 711, optical drive (e.g., CD or DVD) 712, universal serial bus (USB) ports and other communication ports 713, and PCI / PCIe devices 714 may connect to the SB / ICH 702 through bus system 716. PCI / PCIe devices 714 may include Ethernet adapters, add-in cards, and PC cards for notebook computers. ROM 710 may be, for example, a flash basic input / output system (BIOS). The HDD 711 and optical drive 712 can use an integrated drive electronics (IDE) or serial advanced technology attachment (SATA) interface. A super I / O (SIO) device 715 can be connected to the SB / ICH 702.

[0135] An operating system can run on the processing unit 703. The operating system can coordinate and provide control of various components within the data processing system 700. As a client, the operating system can be a commercially available operating system. An object- oriented programming system, such as the Java™ programming system, may run in conjunction with the operating system and provide calls to the operating system from the object-oriented programs or applications executing on the data processing system 700. As a server, the data processing system 700 can be an IBM® eServerTM System® running the Advanced Interactive Executive operating system or the Linux operating system. The data processing system 700 can be a symmetric multiprocessor (SMP) system that can include a plurality of processors in the processing unit 703. Alternatively, a single processor system may¬ be employed.

[0136] Instructions for the operating system, the object-oriented programming system, and applications or programs are located on storage devices, such as the HDD 711, and are loaded into the main memory 704 for execution by the processing unit 703. The processes for embodiments described herein can be performed by the processing unit 703 using computerAttorney Docket No.: 146974.000006 PATENTusable program code, which can be located in a memory such as, for example, main memory 704, ROM 710, or in one or more peripheral devices.

[0137] A bus system 716 can be comprised of one or more buses. The bus system 716 can be implemented using any type of communication fabric or architecture that can provide for a transfer of data between different components or devices attached to the fabric or architecture. A communication unit such as the modem 709 or the network adapter 706 can include one or more devices that can be used to transmit and receive data.

[0138] Those of ordinary skill in the art will appreciate that the hardware depicted in FIG. 7 may vary depending on the implementation. Other internal hardware or peripheral devices, such as flash memory, equivalent non-volatile memory, or optical disk drives may be used in addition to or in place of the hardware depicted. Moreover, the data processing system 700 can take the form of any of a number of different data processing systems, including but not limited to, client computing devices, server computing devices, tablet computers, laptop computers, telephone or other communication devices, personal digital assistants, and the like. Essentially, data processing system 700 can be any known or later developed data processing system without architectural limitation.

[0139] In some embodiments, the data processing system 700 includes or is configured to communicate with a radiation therapy apparatus 720, such as a linear accelerator (LINAC), proton therapy machine, carbon ion machine, heavy ion machine, high electron machine, a FLASH capable device, a device capable of delivering superficial / orthovoltage X-rays, or other medical device capable of delivering therapeutic radiation doses to a subject. The system may establish this communication via multiple interfaces, including direct bus communication 716, the network adapter 706, PCI / PCIe interface 714, or another data connection to facilitate data exchange between the data processing system 700 and the radiation therapy apparatus 720. The radiation therapy apparatus 720 may be integrated with or receive treatment planning data from the data processing system 700 to ensure accurate delivery of the prescribed radiation dose and to adjust parameters dynamically.

[0140] Throughout this disclosure, the following abbreviations may be used: ECOG, Eastern Cooperative Oncology Group; MRI, magnetic resonance imaging; CT (computed tomography); AJCC / UICC: The American Joint Committee on Cancer / The Union for International Cancer Control; LDH: Lactate Dehydrogenase; 2D RT: 2-dimensional radiotherapy; IMRT: intensity modulated radiotherapy, PFS, progression-free survival; UVA, uni variable analysis, MVA, multivariable analysis; RT, radiotherapy; RSI, radio-sensitivityAttorney Docket No.: 146974.000006 PATENTindex; RR, radio-resistant; RS, radio-sensitive; HR, hazard ratio; CI, confidence interval, OS, overall survival.

[0141] Examples

[0142] Example 1: Personalizing Radiotherapy Prescription Dose Using Genomic Markers of Radiosensitivity

[0143] Radiation therapy (RT) is conventionally prescribed based on a uniform, one-size-fits-all approach, delivering small daily doses of RT over several weeks (i.e., fractionation).

[0144] The linear quadratic (LQ) model has been a stalwart in the field that has informed RT dose and fractionation. The LQ proposes that radiation response is a two-parameter function of dose delivered (one parameter, alpha, is linear in dose, and the other, beta, is quadratic). Of note, it has been utilized to calculate equivalent dose and fractionation regimens that have been shown to be safe and effective in clinical trials. However, a fundamental limitation of the LQ model is that it assumes that tumor biology is homogenous and that all individuals in a population have a similar opportunity to benefit from RT, with differences in response being related to probabilistic events. Thus, the LQ model predicted that uniform RT dose escalation would result in significant clinical gains across multiple disease sites. However, multiple prospective Phase 3 randomized trials have disproven this prediction.

[0145] The development of “omic” technologies has revealed that cancer is the most heterogeneous and complex disease that affects humans. Rather than a single disease with a uniform treatment, the complexity and diversity of cancer requires many treatment options that are matched and optimized based on the patient’s individual tumor biology.

[0146] Although RT remains a critical curative agent for cancer, it has yet to adapt a biological basis in the clinic. It was previously proposed that the gene expression-based radiosensitivity index (RSI), a surrogate for intrinsic cellular radiosensitivity, and the genomic- adjusted radiation dose (GARD), an individualized quantitative metric of the clinical effect of RT, could serve as the first approach to biology-based RT. Both RSI and GARD have been validated as a predictor of clinical outcome in patients treated with RT.

[0147] RSI is derived from tire expression levels of specific genes, introducing variability from both biological sources - such as tumor heterogeneity and genetic polymorphisms - and technical factors, including the efficiency of RNA extraction and the accuracy of gene expression profiling. This variability is reflected in the standard deviation of RSI, which has been observed to be approximately 10%. Furthermore, reproducibility studies demonstrated a median RSI difference of 0.06 (6.47% of range) across samples, with some samples showingAttorney Docket No.: 146974.000006 PATENTvariability up to 15%. This variability directly affects the calculation of the patient-specific radiosensitivity parameter (a) used in the LQ model, thereby impacting the precision of GARD.

[0148] RSI is calculated generally with the equation:RSI=k1*AR+k2*c-jun+k3*STAT1+k4*PKC+k5*RelA+k6*cAbl+k7*SUMO1+k8*PAK2+k9*HDAC+k10*IRF1and specifically using the following equation with constants ki-ke substituted:RSI=-0·0098009*AR + 0·0128283*cJun + 0.0254552*STAT1 - 0.0017589*PKC - 0·0038171*RelA + 0·1070213*cABL - 0.0002509*SUMO1 - 0·0092431*PAK2 - 0·0204469*HDAC1 - 0·0441683*IRF1

[0149] RSI is based on gene expression data, is not cancer-type specific, and can be used across a range of malignancies.

[0150] GARD is a personalized metric designed to quantify the expected biological effect of RT based on an individual’s genomic profile. GARD integrates the patient-specific RSI with tire LQ model, creating a tailored approach to radiation dosing. However, there can be variability in the GARD calculation, largely due to the processes involved in determining RSI.

[0151] GARD is derived using the LQ model, the individual RSI and the radiation dose / fractionation schedule for each patient. First, a patient-specific a is derived by substituting RSI for Survival (S) in the LQ equation below where dose (d) is 2Gy, n = 1 and P is a constant (0.05 / Gy2):S=e’”d(a+|5f])

[0152] GARD is calculated using the classic equation for biologic effect, GARD = nd (a + (3d), and using the patient-specific a is calculated as stated above, and the number of fractions (n) and dose per fraction (d) received by each patient.

[0153] RxRSI is the physical dose required to achieve a previously identified GARD threshold. RxRSI is calculated using the following formula:RxRSI = GARD target / (a + [3d)where a is calculated based on the patient’s RSI as described above and [3 is a constant (0.05 / Gy2).

[0154] Additionally, the steps involved in processing the sample, isolating RNA, and radiating the chip each contribute to the overall uncertainty in RSI, which subsequently influences GARD. The observed variability in RSI, as well as in other gene expression signatures, underscores the range of possible GARD values rather than a single, precise figure.Attorney Docket No.: 146974.000006 PATENT

[0155] This inherent variability justifies the need to interpret GARD within a range when considering its application. Therefore, when reference is made to “about” a certain GARD value, it is understood to mean within a range of + / - 10%, providing the necessary flexibility to account for these technical variations. The flexibility allows the GARD values to be interpreted with a margin that accommodates not only the median variability observed but also the higher variability seen in certain samples, which can extend beyond the 6% range to as much as 15%> in some cases.

[0156] The distribution of both the RSI and the GARD varies significantly across different disease sites, reflecting inherent biological variability. For example, cancers such as gliomas, soft tissue sarcomas, and melanomas tend to exhibit more radioresistant profiles, with higher mean RSI values. In contrast, oropharyngeal cancer typically displays more radiosensitive distributions, characterized by lower mean RSI values. This variability underscores the necessity of disease specific calibration of GARD, as the underlying radiosensitivity and therapeutic responses to RT differ markedly across cancer types.

[0157] Furthermore, GARD, which quantifies the therapeutic benefit of RT, must account for these inter-disease differences. The therapeutic efficacy of RT is not uniform across all cancers: rather it is highly dependent on the biological characteristics of each disease site. For instance, in oropharyngeal cancers, RT is often employed as the primary curative modality, necessitating precise calibration of GARD to reflect its therapeutic role. The decision to assess oropharyngeal cancers in the below Examples 2 and 3 is thus based on their distinct radiosensitivity profiles and the pivotal role of RT in their treatment, which requires tailored GARD calibration to ensure accurate therapeutic predictions.

[0158] Example 2: Evaluation of HPV-Positive Oropharyngeal Cancer Using GARD: Analysis of RT Outcomes and Potential for Treatment De-escalation

[0159] Treatment decision-making in oropharyngeal squamous cell carcinoma (OPSCC) includes clinical stage, HPV status, and smoking history. Despite improvements in staging with separation of HPV positive and negative OPSCC in AJCC 8th edition (AJCC8), patients are largely treated with a uniform approach, with recent efforts on de-intensification in low-risk patients. GARD is shown that it can be used to predict overall survival (OS) in HPV-positive OPSCC patients treated with RT.

[0160] Methods

[0161] A total of 1,537 patients (1,086 retrospectively and 451 prospectively) with loco- regional advanced head and neck cancer (Stage III-IVa, IVb) treated with curative intent, including 377 patients with HPV+ oropharyngeal cancer. After excluding patients withAttorney Docket No.: 146974.000006 PATENTinadequate tissue sample (n=85), poor RNA quality (n=6), HPV DNA-negative (n=14), patients that underwent single-modality treatment (n=37) and one patient treated with surgery alone, a final study population of 234 patients remained. Patients were treated in a nine-year window, with follow-up closed two years thereafter. HPV testing was performed with p 16 immunohistochemistry and confirmed by HPV DNA testing following positive staining. In total, 191 patients received definitive RT primary treatment (chemoradiation (n=172) or RT alone (n= 19)), and 43 patients received post-operative RT (post-op chemoradiation (n=29) or post-op RT alone (n=14)). Two RT dose fractionations were utilized for definitive RT cases (70 Gy in 35 fractions or 69.96 Gy in 33 fractions). Median RT dose was 70 Gy (mean 50.88-74) for definitive cases and 66 Gy (range 44-70) for post-operative cases. The median follow up was 46.2 months (IQR 33.5-63.1).

[0162] Gene expression profiles (Affymetrix Ciariom D) were analyzed for 234 formalin-fixed paraffin-embedded samples from HPV-positive OPSCC patients within an international, multi-institutional, prospective / retrospective observational study. Ill patients were stage I, 64 were stage II, and 59 were stage III. GARD was calculated for each patient as previously described. In total, 191 patients received definitive treatment (chemoradiation or RT alone), and 31 patients received post-operative RT. Two RT dose fractionations were utilized for definitive cases (70 Gy in 35 fractions or 69.96 Gy in 33 fractions). Median RT dose was 70 Gy (mean 50.88-74) for definitive cases and 66 Gy (range 44-70) for post-operative cases. 29 patients received surgery and adjuvant chemoradiation, 14 received surgery and adjuvant RT, and 19 received RT alone. The median follow up was 46.2 months (95% CI, 33.5-63.1). Cox proportional hazards analyses were performed with GARD as a continuous variable and ROC analyses compared the performance of GARD with AJCC8.

[0163] All patient tumors underwent gene expression profiling using Affymetrix Ciariom D on their formalin fixed samples and RSI values were generated using a 10-gene signature. A patient specific genomic parameter, ag, was subsequently calculated using the linear-quadratic model to estimate patient radiosensitivity using the relation:ln7?57 „,U,> — —. — I5u,ndwhere dose d is 2 Gy, n is 1. The assumption is made that |3 is a constant at 0.05 / Gy2. This genomic agis then used together with each patient’s specific radiation dosing to calculate their clinical GARD value (GARDc):GARDc = ncdc(αg+ βdc),Attorney Docket No.: 146974.000006 PATENTwhere ncis the number of fractions and dcis the dose per fraction per the clinically delivered radiation plan to each patient. These values were calculated without information about clinical outcome. Cox proportional hazards regression was used to assess the association between GARD as a continuous variable and OS. OS was defined as the time between primary diagnosis and death or last follow-up. While the continuous analysis is statistically the most rigorous, to make clinical translation simpler, a series of discrete analyses was performed to mimic the discrete dose levels preferred by clinicians. Discrete analyses were performed with median GARD and also, for hypothesis generation, using an algorithm to minimize chi-squared to derive presumed optimal groups in three dose levels.

[0164] Each virtual patient was randomly selected from the RSI distribution of the complete cohort. The virtual patients were randomized to receive 70Gy in 35 fractions or 60Gy in 30 fractions. 200 patients per arm were used. GARD was calculated for each virtual patient, and an OS curve was predicted based on the GARD level achieved, using the optimized three dose GARD level approach (GARD < 46.1 = low, 46.1< GARD < 65 = intermediate, GARD > 65= high). Each of these survival curves was modeled with a Weibull curve fit to the Kaplan-Meier estimate of that GARD level. The weighted average of individual patients’ survival curves represented the overall survival estimate for each trial arm. This entire process was repeated a total of 20 times to replicate the variability between different groups of patients.

[0165] To determine whether GARD could identify a successful de-intensification strategy, a variation of this trial with selective de-intensification based on GARD was also performed. In this approach, rather than de-intensifying all patients, only patients who would remain in either GARD dose level high (GARD > 65= high) or intermediate (46.1< GARD < 65 = intermediate) with either 60 or 70 Gy were assigned to 60 Gy. All patients that achieve GARD dose level low (GARD < 46.1 = low) were excluded from selective de-escalation. In addition, patients who drop from GARD high to GARD intermediate when de-escalated were also excluded from selective de-escalation.

[0166] After performing standard analyses to determine the association of GARD with outcome, it was sought to incorporate GARD into a model using previously validated variables. Subsequently, a nomogram incorporating GARD, AJCC8 staging, the three-cluster gene expression model, and other clinical variables was constructed following standard methods. The nomogram was validated by comparison of receiver operating characteristic (ROC) curve analysis.

[0167] ResultsAttorney Docket No.: 146974.000006 PATENT

[0168] Despite uniform radiation dose utilization, GARD showed significant heterogeneity (range 30-110), reflecting the underlying genomic differences in the cohort. On multivariate analysis, each unit increase in GARD was associated with an improvement in OS (HR = 0.9539, (0.9140, 0.9955), p = 0.0306) compared to AJCC8 (HR = 2.3150, (0.9260, 5.7875), p = 0.0726). ROC analysis for GARD at 36 months yielded an AUC of 81.6 (70.8, 92.4) compared with an AUC of 65.0 (48.9, 81.0) for the NRG clinical nomogram. GARD>64.2 was associated with improved OS (HR = 0.280 (0.100, 0.781), p = 0.015). In this virtual trial, GARD predicts that uniform RT dose de-escalation results in overall inferior OS but proposes two separate genomic strategies where selective RT dose de-escalation in GARD-selected populations results in clinical equipoise.

[0169] As shown in FIG. 1A, GARD ranged from 31.7 to 108.9 (IQR56.4-71.9) with a median for the whole cohort of 63.5. Plotted along the edges of the joint plot between GARD and EQD2 are kernel density estimates for die entire cohort, revealing wide heterogeneity in GARD (std = 13.8) in the setting of near homogeneity in RT dosing (std = 3.1). The difference between GARD and EQD2 is best exemplified by the patients treated with ‘standard’ fractionation - with EQD2 measures between 69-71, see FIG. IB. The range of GARD for those patients ranged was 33.7-108.9 (IQR 56.9-72.9) even though they all were treated to (approximately) the same RT dose (EQD2), highlighting the wide differential in the predicted effect of uniform clinical dosing strategies. GARD is continuously associated with OS in RT-treated HPV-positive OPSCC patients. Previously, it has been demonstrated that GARD was associated with overall survival, recurrence risk and was predictive of RT benefit in a pooled pan-cancer analysis including 1615 patients. RT therapeutic benefit is a critical factor impacting clinical outcome in HPV+ patients. It is shown herein that GARD would be associated with clinical outcome in this analysis of HPV+ oropharyngeal squamous cell carcinoma patients collected through the B2DECIDE project.

[0170] As seen in FIGs. 1A-1B, GARD exhibits large underlying genomic heterogeneity in radiation effect compared to radiation dose alone. In FIG. 1A, EQD2 (median 70.0, std 3.1) is plotted against associated GARD (median 63.5, std 13.8) for each patient in the whole cohort. Kernel density estimates are plotted on each edge to show the distributions of the individual variables. As seen in FIG. IB, plotting patients who received an EQD2 of 69-71 Gy only (standard dosing) highlights GARD’s ability to stratify patients by their genomic heterogeneity. Data points are overlaid with a box-whisker plot with box representing quartiles and whiskers extending to 1.5 times IQR.Attorney Docket No.: 146974.000006 PATENT

[0171] GARD was the only variable statistically associated with OS both as a continuous (HR = 0.954 (0.914,0.996), p = 0.031) and discrete variable (HR 0.291 (0.103,0.818), p = 0.019) (Table 1). Smoking (pack-years) was associated with OS as a continuous variable but not as a discrete variable. To further evaluate GARD’s prognostic ability a discrete analysis using the GARD median as a cut-point (GARD = 64.2.) was performed for the patients treated with primary definitive RT. It was found that GARD high vs. low significantly associated with OS (p = 0.01) (FIG. 3A): GARD-high patients had a 99.0% and 94.6% 3-year and 5 year-OS rates, whereas GARD-low patients achieved an 89.2% and 79.2% 3 year and 5-year OS rate, respectively. This dichotomization resulted in a HR of 0.280 (0.100, 0.781), p = 0.015.

[0172] Table 1. Multivariate analysis of definitive RT patientsHR (95% CI) PGARD (continuous) 0.950 (0.909, 0.992) 0.020Stage - high 2.050 (0.810, 5.190) 0.130ECOG = 1 or 2 0.778 (0.221, 2.737) 0.696Pack years (continuous) 1.023 (1.004, 1.042) 0.019GARD (discrete) 0.230 (0.079, 0.672) 0.007Stage = high 1.819 (0.718, 4.608) 0.207ECOG = 1 or 2 0.727 (0.206, 2.573) 0.621Pack years (continuous) 1.029 (1.007, 1.05) 0.008GARD (continuous) 0.950 (0.910, 0.991) 0.019Stage = high 2.095 (0.831, 5.281) 0.117ECOG = 1 or 2 0.788 (0.224, 2.769) 0.711Pack years (discrete) 2.098 (0.811, 5.427) 0.127GARD (discrete) 0.257 (0.090, 0.731) 0.011Stage = high 1.895 (0.753, 4.769) 0.175ECOG = 1 or 2 0.789 (0.225, 2.765) 0.711Pack years (discrete) 2.264 (0.875, 5.855) 0.092

[0173] Having shown in previous analyses that GARD was predictive of outcome in pooled cohorts, it is now to be demonstrated that it would be continuously associated with OS in this, the largest cohort of OPHNSCC with genomics, details on radiation treatment and clinical outcome gathered to date.

[0174] To test this, a Cox proportional hazards analysis of GARD and OS was performed in both the entire cohort, and also the subset that was treated with RT alone. As shown in FIGs.2A-2B, GARD is associated with OS as a continuous variable in both groups. For each unit increase in GARD there is an improvement in OS (HR = 0.967 (0.937, 0.998) per unit GARD, p = 0.038) in the entire cohort. This association of GARD with OS was statistically stronger when including only patients treated with definitive RT (HR = 0.955 (0.915, 0.996) per unit GARD, p = 0.038). This association of GARD with OS was statistically stronger whenAttorney Docket No.: 146974.000006 PATENTincluding only patients treated with definitive RT (HR = 0.955 (0.915, 0.996) per unit GARD, p = 0.030). Examining only patients who received the standard-of-care range 69-71 Gy EQD2 within this group, the HR was 0.940 (0.899, 0.983), p = 0.007. While significant differences throughout the entire cohort were found, to keep the cohort population uniform, the remainder of the analysis focused on definitive primary RT patients.

[0175] As seen in FIGs 2A-2B, GARD was a continuous predictor of OS in radiation treated patients with HPV+ OPHNSCC. Performing a Cox regression analysis for GARD as a continuous variable revealed statistically significant associations with OS for the entire cohort (FIG. 2A), and an even stronger signal for patients treated with RT alone (FIG. 2B). FIG. 2A shows Cox proportional hazards analysis demonstrates significant continuous association between GARD and OS for the entire cohort (p = 0.038, HR = 0.967 (0.937, 0.998) per unit GARD). FIG. 2B shows Cox proportional hazards analysis demonstrates significant continuous association between GARD and OS for the subset of patients treated with RT alone (p = 0.030, HR = 0.955 (0.915, 0.996) per unit GARD).

[0176] The data in FIG. 2A are stratified by median GARD, while the data in FIG. 2B are organized in optimal tertiles.

[0177] To further evaluate GARD’s prognostic ability, a discrete analysis using the GARD median as a cut-point (GARD=64.2) is performed in the patients treated with definitive RT. FIG. 3A shows that GARD-high patients have an improved OS when compared with GARD-low patients (p = 0.01). GARD-high patients have a 99% and 94.6% 3-year and 5 -year OS rates whereas GARD-low patients achieve an 89.2% and 79.2% 3 year and 5-year OS rate, respectively. This dichotomization results in a HR of 0.28 (0.10, 0.78) with p = 0.015.

[0178] FIGs. 3A-3B show discrete GARD cutpoints display survival differences between groups. FIG. 3A shows stratifying patients by median GARD shows a significant difference in OS at 5 years by the log-rank rest. To generate hypotheses regarding poorest performing groups, the cohorts are stratified by two cutpoints by minimizing the chi-squared statistic and calculating the log-rank statistic as seen in FIG. 3B.

[0179] To read this nomogram shown in FIG. 4, tally points from the first row for a patient’s corresponding GARD, AJCC8 stage, and molecular cluster group. The total points correspond to a 3-year survival estimate.

[0180] To determine whether GARD could identify a group of patients with poor prognosis, an exploratory discrete analysis based on an optimized two cut-point analysis is performed. Minimizing the chi-square at two discrete values reveals three groups with maximally different outcomes, as shown in FIG. 3B. This analysis revealed two cutpoints at GARD 65 and 46.1.Attorney Docket No.: 146974.000006 PATENTPatients that achieve the highest GARD (GARD > 65) have a 3year OS of 100% compared with 91.3% (85.3, 97.8) for the GARD intermediate group (46.1< GARD < 65) and 62.5% (33.6, 100) for the group that achieves the lowest GARD (GARD < 46). These differences are statistically significant with p < 0.001, though this statistic be interpreted carefully as the groups were chosen by maximizing differences post-hoc.

[0181] To whether empiric dose de-escalation results in a small number of patients failing from the GARD intermediate cohort (46.1< GARD < 65) to the GARD low cohort (GARD < 46) leading to an inferior result for empiric dose de-escalation, an in silico clinical trial was performed to evaluate GARD-based predictions of clinical outcome for empiric dose de-escalation to 60 Gy (with concurrent chemotherapy). It was found that GARD predicts that empiric dose de-escalation would result in an inferior clinical outcome. The predicted 3-year OS for patients modeled at 70 Gy is 94.2% compared with 90.2% for patients modeled at 60 Gy (FIG. 6A). Empiric, unselected dose de-intensification is predicted to increase the proportion of patients in the GARD low group while decreasing the proportion of patients in the GARD high group. The 70Gy in silico arm had 13, 90, and 97 patients in the low, intermediate, and high GARD groups, while the 60Gy in silico arm had 36, 123, and 41 patients in those groups.

[0182] To further evaluate the ability of GARD to identify subpopulations at higher risk of failure, a re-analysis of the GARD cutpoint was conducted. Although HPV-positive patients generally have an excellent prognosis, their remains a need for more precise tools to identify patient subsets who may not benefit from uniform treatment de-escalation. It was hypothesized that GARD could serve this purpose. Given these considerations, a discrete analysis was performed to determine an optimized GARD cutpoint, which indicated 41.97 as a critical threshold for stratifying patients based on overall survival outcomes.

[0183] FIGs. 3C-3D depict the Kaplan-Meier survival curves for both RSI (FIG. 3C) and GARD (FIG. 3D) using this updated GARD cutpoint of 41,97. In the RSI plot, the “rsi„optimal=lc” line represents patients with RSI values below the threshold, indicating lower radiosensitivity, while the “rsi..optimal=high” line represents those above the threshold, indicating higher radiosensitivity. The GARD plot shows that patients with GARD values greater than 41.97 (“gard_optimal=high”) have markedly beter survival outcomes compared to those with lower GARD values (“gard. optimal=low”). The separation between the curves, particularly in the GARD plot (p = 0.0045), underscores the utility of GARD in stratifying patient risk more effectively than RSI alone. These findings further support the use of GARDAttorney Docket No.: 146974.000006 PATENTin identifying patients who may not be suitable candidates for treatment de-escalation and in guiding personalized RT decisions in this patient population.

[0184] Next, it was determined that GARD can be used to develop a clinical trial strategy that predicts equivalent outcome at 70 or 60 Gy. In one approach, GARD can identify patients that would remain above the GARD-high cutpoint (65) at 70 or 60 Gy. This approach selects patients with RSI< 0.115 which compromise 18% of the total HPV+ population. In another approach, GARD-high and intermediate patients at 70 Gy that remain in the same group at 60 Gy, would also be predicted to achieve equipoise with selective dose de-escalation. Approximately 55% of HPV+ patients would be eligible for this approach. It should be noted that both approaches exclude GARD-low patients (as these patients are predicted to require dose intensification) and patients that fall from GARD-high to GARD-intermediate or GARD-intermediate to GARD-low at 60Gy. The predicted OS curve for the second approach to de-escalation is shown in FIG. 6B.

[0185] GARD outperforms AJCC8 as a prognostic factor in HPV -positive oropharyngeal patients The Cox analysis suggested that GARD may outperform TNM8 as a prognostic factor in this group of patients. To further test this, ROC analysis comparing GARD and TNM8 as prognostic factors for OS (3-year) is performed. FIG. 5 shows that GARD achieves an AUC = 81.6 compared to an AUC - 65.0 for TNM8, consistent with GARD being more accurate model for OS than standard clinical stage.

[0186] A clinical nomogram including staging and genomics significantly outperforms AJCC8 To develop a more accurate model of individualized risk, a clinico-genomic nomogram was constructed incorporating GARD, TNM8, and the previously developed 3-cluster prognostic model FIG. 5. The nomogram awards points based on each individual patient’s GARD, clinical stage and cluster prognostic group. A vertical line is drawn from the total points calculated to the predicted 3-year OS for the patient. When the cohort was assessed for prognosis via the nomogram the total points calculated ranged from 8.1-184.7 which corresponded to predicted 3-year OS from 70.04-99.79. The performance for the nomogram is shown in Table 2.

[0187] Table 2.3-year AUC (95%CI)AJCC8 65.0 (48.9, 81.0)Clusters 72.8 (59.0, 86.7)GARD 81.6 (70.8, 92.4)AJCC8 / GARD / clusters 84.0 (70.6, 97.5)Attorney Docket No.: 146974.000006 PATENT

[0188] Both GARD and the 3-cluster prognostic model outperform TNM8 as a single parameter model (GARD vs. 3-cluster vs. TNM8 AUC (3-year OS) 81.6 vs 72.8 vs. 65.0). The combination of GARD / 3-cluster / TNM8 achieved the best performance (AUC 84.0). FIGs. 6A-6B show that GARD predicts the results of treatment de-intensification. FIG. 6A shows that uniform de-intensification produces decreased OS. FIG. 6B shows that selective deintensification produces similar OS even when approximately two-thirds of patients in the second arm receive the lower RT dose.

[0189] Discussion

[0190] The development of prognostic models to more accurately classify cancer is a central goal of personalized medicine. It is shown herein that GARD, a previously generated model of the treatment effect of RT, is associated with overall survival in HPV+ oropharyngeal cancer patients treated with RT both as a continuous and dichotomous variable. Furthermore, using time-dependent ROC analysis, it is shown that GARD outperforms the current clinical nomogram for prediction of overall survival of these patients. Finally, it is shown that GARD predicts that uniform RT dose de-escalation would result in an inferior overall survival over standard RT dose. However, GARD proposes two different clinical strategies to selective RT dose de-escalation that it predicts would achieve clinical equipoise.

[0191] Since its confirmation as a biomarker of outcome, HPV status has been incorporated into the diagnostic algorithm of the disease. In this analysis, it is demonstrated that GARD identifies a 72% reduction in the risk of death in GARD-high HPV+ patients. This translates into an absolute 10% and 16% difference in 3- and 5-year OS between GARD-high and GARD- low patients. In addition, GARD outperformed the established nomogram for overall survival. Finally, a model integrating GARD and the nomogram achieved the highest prognostic ability. Thus, GARD can resolve prognosis for HPV+ patients with the same magnitude that HPV did for head and neck cancer patients.

[0192] A central clinical question for HPV+ oropharyngeal cancer patients is whether their treatment can be deintensified while preserving their excellent prognosis. Clinical factors alone are not enough to identify patients where radiation dose de-intensification can be performed without clinical outcome detriment.

[0193] While it is demonstrated that GARD outperforms standard clinical variables as a prognostic factor in HPV+ oropharyngeal cancer, GARD can further improve the ability to define appropriate subpopulations for treatment de-intensification. For example, in an exploratory analysis, it is shown that GARD identifies a small group of HPV+ patients with poor prognosis (GARD < 46.1) who achieve a 3-year OS of 62.5%. In addition, GARD is alsoAttorney Docket No.: 146974.000006 PATENTan actionable model that can provide guidance on RT dosing for genomically defined subpopulations. This can inform the design of the next generation of clinical trials for these patients. Proof of principle is demonstrated for GARD-based clinical trial design by showing that GARD predicts that empiric dose de-escalation would result in an inferior 3-year OS for the patients treated to the lower dose. Finally, at least two designs are proposed that show that GARD-based modeling predicts would result in clinical equipoise between 70 and 60 Gy with appropriately chosen patients for de-escalation. In the first design, only patients in the top 18% of the GARD distribution would be eligible, while in the second design approximately 55% of HPV+ patients would be eligible. A key observation is that a small subset of patients may need dose intensification and should not be eligible for these trials.

[0194] It is demonstrated that GARD outperforms the clinical nomogram of outcome as a prognostic biomarker in HPV+ OPSCC patients and defines prognostic groups that can inform clinical trial design. While HPV is a classic biomarker in that its result is fixed and cannot be changed, the GARD value for a patient can be optimized by adjusting the RT dose. This supports the hypothesis that GARD could be used to optimize clinical outcome for HPV+ oropharyngeal SCC patients by the personalization of RT dose. Even without the use for dose personalization, however, the strong improvement (quantitatively equivalent to the seminal findings of HPV positivity itself) in outcome prognostication suggests that obtaining GARD should be considered for HPV+ oropharyngeal cancer patients.

[0195] Example 3: Personalization of Radiotherapy Dose in HPV Positive Oropharynx Cancer Using GARD

[0196] As discussed above, since the discovery that HPV is an etiologic and strong prognostic factor in oropharyngeal squamous cell carcinoma, assessing this biomarker indirectly via pl6 or directly via in situ hybridization has become standard of care in the diagnostic and staging work up of these patients. A three-group classification system based on clinical factors (HPV status, pack years of smoking, and T or N classification) has informed the design of multiple clinical trials. As the low-risk group in this classification had an OS of 93% at three years, it was hypothesized that this favorable subset could be treated to a lower RT dose / toxicity without detriment in OS.

[0197] While several approaches to RT dose de-intensification have been explored, all of them share a number of characteristics. First, they define eligibility based on clinical factors that define prognostic risk. Second, they assume all patients are biologically homogeneous and have the same opportunity to benefit from RT. Therefore, the RT de-escalation strategy utilized isAttorney Docket No.: 146974.000006 PATENTuniform. Some studies chose 60 Gy with concurrent cisplatin, while others chose uniform 30 Gy to the neck after surgery with no chemotherapy.

[0198] Interim results in a Phase 3 clinical trial testing the non-inferiority of uniform RT dose de-escalation (cisplatin + 60 Gy or nivolumab + 60 Gy) against the standard of care (cisplatin + 70 Gy) failed to demonstrate the non-inferiority of cisplatin + 60 Gy over standard of care. These results suggest that clinical factors and a uniform therapeutic approach are not enough to provide therapeutic guidance for RT de-intensification and suggest that similar' to many targeted and immunotherapy agents, RT dose optimization may need to be targeted to specific genomically- defined subpopulations.

[0199] In previous studies, a GARD, a radiation-specific metric that was shown to quantify the RT treatment effect in a given patient as a function of their RT dose and tumor genomics. GARD results suggest that the treatment effect of a uniform dose of RT (e.g., 70 Gy) is biologically highly heterogeneous, rather than homogenous, the current assumption in the field. In a pooled analysis of 1,615 patients in seven different disease sites, it was demonstrated that GARD was associated with overall survival and recurrence risk as a continuous variable and predicted RT treatment benefit for each individual patient. Since GARD quantifies the treatment benefit for each individual patient, GARD-based models can be used to inform RT dose adjustments to optimize a patient’s clinical outcome.

[0200] GARD provides a critical innovation compared with the current approaches to RT deintensification: the ability to depart from the assumption that RT benefit is homogenous and the limitation of uniform RT dosing strategies. Since RT benefit is one of the critical factors defining clinical outcome in HPV -positive HNSCC patients, it was hypothesized that GARD could provide novel information that will allow for better more personalized approaches to the successful treatment of these patients.

[0201] To test whether GARD could serve this purpose, its prognostic ability was assessed in a cohort of radiation treated patients with HPV positive HNSCC. Further, it was hypothesized that GARD’s prognostic information would provide an improvement to outcome prediction compared to clinical factors alone. Since GARD is intrinsically linked to radiation dose, any improvement in outcome prediction utilizing GARD can also fundamentally be used to make quantitative predictions for differential outcome given specific RT dose adjustments.

[0202] This Example 3 details an analysis of patients treated with radiation therapy with HPV- positive HSNCC. Individual patient RSIs were assessed from gene expression data derived from formalin fixed tumor specimens, and radiation-dosing information was used for each patient to calculate GARD. Continuous Cox proportional hazards regression was used to determine theAttorney Docket No.: 146974.000006 PATENTrelationship between GARD and outcome, and a discrete analysis at several cutpoints was presented post hoc to suggest optimal stratification strategies. GARD-based models were then developed to personalize RT dose to achieve the best possible clinical outcome (both tumor and normal tissue) for each individual patient. The results indicates that it is not only possible to reduce RT dose for a significant number of patients but that it is also possible to improve tumor outcomes with RT dose personalization.

[0203] Methods

[0204] A total of 1,537 patients (1,086 retrospectively and 451 prospectively) were enrolled with loco-regional advanced head and neck cancer (Stage Ill-lVa, IVb, AJCC 7th edition, or Stage I - III, AJCC 8th edition) treated with curative intent, including 377 patients with HPV-positive oropharyngeal cancer. Of these, 286 patients had gene expression profiling available.

[0205] After excluding patients that were HPV DNA-negative (n=14), patients with locally-advanced disease that underwent single-modality treatment (n=37), patients treated with surgery alone (no GARD could be calculated) (n=1), and patients with post-operative RT (n=43), a final study population of 191 patients remained.

[0206] HPV testing was performed with p!6 immunohistochemistry and confirmed by HPV DNA testing following positive staining. In total, 191 patients received definitive RT primary treatment. FIG. 8 provides details on patient selection. Fifteen RT dose fractionations were prescribed for definitive RT cases with the two most common being 70 Gy in 35 fractions or 69.96 Gy in 33 fractions. Median RT dose was 70 Gy (range 51-74) for definitive RT cases. The median follow up was 43.95 months (IQR 33.0-60.7).

[0207] All patient tumors previously underwent gene expression profiling using Affymetrix Clariom D from formalin fixed samples. Raw CEL files were processed and normalized using Affymetrix sst-rma in Expression Console (Thermo Fisher ), then RSI values were generated using a 10-gene signature, as implemented in the R package hacksig. RSI has been previously clinically validated in multiple cohorts. A patient-specific genomic parameter, αg, was subsequently calculated using the linear-quadratic model to estimate patient radiosensitivity, the derivation of which we have previously described, yielding Hie relation:InRSIn,a9n— - ndfBa,

[0208] where dose d is 2 Gy, and the number of fractions, n is 1, as this moves from a genomic measure to the familiar Surviving Fraction after 2Gy (SF2). is a constant at 0.05 / Gy2. This genomic parameter, αg, is then used together with each patient’s specific radiation dose and fractionation to calculate their clinical GARD value (GARDc):Attorney Docket No.: 146974.000006 PATENTGARDc= ncdc(αg+βdc),

[0209] where ncis the number of fractions and dcis the dose per fraction per the clinically delivered radiation plan to each patient. These values were calculated without information about clinical outcome.

[0210] OS comparisons were performed. Cox proportional hazards regression was used to assess the association between GARD as a continuous variable and OS. Multivariable Cox regression analysis was performed. Survival curves were generated. Discrete analyses with log¬ rank statistics were performed for hypothesis generation, using an algorithm to minimize the logrank statistic to derive optimal groups in two dose levels.

[0211] OS was defined as the time between primary diagnosis and death or last follow-up. Follow up was censored at 60 months if no event occurred prior. After performing standard analyses to determine the association of GARD with outcome, GARD was incorporated with known prognostic clinical variables including stage, pack-years smoking and ECOG performance status to determine whether GARD improved prognostic performance. In addition, the three-cluster gene expression model was integrated following standard methods. The Cox proportional hazard models were evaluated by comparison of time-dependent ROC curve analysis.

[0212] An in-silico trial to test uniform RT dose de-intensiflcation (from 70Gy to 60Gy) was performed in this cohort of HPV-positive oropharyngeal cancer patients. Virtual patients were generated by sampling from the density estimates of the empirical RSI distribution. Outcomes are estimated based on a Weibull distribution of the original outcomes evaluated at discrete timepoints. The combination of low / high GARD population distributions yield the overall cohort outcome estimates. GARD was calculated for each virtual patient, and an OS curve was predicted based on the GARD level achieved, using the optimized two dose GARD level approach (GARD < 42 = low and GARD > 42 = high). This cutpoint is identified as 41.97 in FIG. 9.

[0213] To determine whether GARD could identify a successful de-intensification strategy, a two arm in silico trial was performed in which only a GARD-identified subset of patients was de-intensified in one arm. The control arm was treated uniformly at 70 Gy, as above. In the in silico experimental arm, only GARD-high patients (GARD > 42) who remained in the same risk group after de-escalation to 60 Gy were de-intensified. All other patients remained at 70 Gy. In other words, GARD-low (high-risk) patients are treated at 70 Gy, and GARD-high patients are de-intensified to 60 Gy unless this changes their risk category.

[0214] It was also sought to determine what would be revealed by a purely personalized approach, with a GARD target chosen to provide an overall outcome equivalent to the currentAttorney Docket No.: 146974.000006 PATENTstandard of care. In this personalized iso-curative dosing strategy, a standard of care arm (70 Gy in 35 fx) was compared to a target GARD identified to provide equipoise to modern outcomes (in this case GARD = 32), see FIG. 10. The total difference in dose predicted to provide equipoise across the population, and the differential in cost to provide this at the population level, was then calculated.

[0215] Results

[0216] 191 patient tumors meeting the criteria of definitive RT were identified in the cohort. FIG. 8 provides details on patient selection. The characteristics for these HPV -positive oropharyngeal squamous cell carcinoma patients are detailed in Table 3, below. Counts are provided in parentheses except for Smoking Pack Years, which is reported as median and interquartile range (IQR).

[0217] Table 3.Characteristic N = 191AJCC8, n (%)I 87 (46)II 49 (26)III 55 (29)ECOG, n (%)0 158 (83)1 32 (17)2 1 (0.5)Smoking Pack Years, Median (IQR) 8 (0 - 30)AJCC8 T stage, n (%)T1 38 (20)T2 57 (30)T3 45 (24)T4 51 (27)AJCC8 N stage, n (%)NO 6 (3.1)N1 143 (75)N2 34 (18)N3 8 (4.2)Status, n (%)Alive 172 (90)Dead 19 (9.9)0218] It has been previously shown that GARD reveals underlying heterogeneity in radiation treatment effect within groups presumed to have been treated uniformly (with approximately equivalent physical dose). In this cohort, it is again demonstrated that GARD reveals wide heterogeneity in predicted RT effect in spite of relatively uniform RT dose prescribed. As shown on the left side of FIG. 11, delivered GARD ranged from 15.4 to 71.7 (median: 39.1, IQR: 12.6). Plotted along the edges of the jointplot between GARD and EQD2 are kernel density estimates forAttorney Docket No.: 146974.000006 PATENTthe entire cohort, revealing wide heterogeneity in delivered GARD (IQR 12.6 ) in the setting of near homogeneity in RT dose (IQR 0.04). The difference between GARD and EQD2 is best exemplified by the patients who received the whole course of ‘standard’ radiation dose - with EQD2 measures between 69-71 Gy, see the right side of FIG. 11. The range of GARD for those patients was 19.7-71.7 (IQR 12.7) even though they all were treated to (approximately) the same RT dose (EQD2), highlighting the wide differential in the predicted effect of the uniform clinical dosing strategies. The distributions of RSI and GARD by AJCC8 stage did not differ significantly. See FIG. 15 and Table 4, below.

[0219] Table 4.Stage N Median (IQR) PRSI 1 87 0.32 (0.28 - 0.230.38)II 49 0.33 (0.26 - 0.39)III 55 0.35 (0.28 -- 0.41)GARD I 87 40 (34-46) 0.25II 49 39 (34-47)III 55 37 (31 -45)

[0220] To test whether GARD would be associated with clinical outcome in this analysis of HPV-positive oropharyngeal squamous cell carcinoma patients, a Cox proportional hazards analysis of GARD and OS in patients that were treated with definitive primary RT (n=191) and those treated with standard of care definitive primary RT (EQD2 69-71 Gy) (n=174) was performed. As shown in FIG. 12A, GARD is associated with OS as a continuous variable for patients treated with primary definitive RT and censored at 60 months. For each unit increase in GARD, it was found that there is an improvement in OS (HR (95% Cl) = 0.941 (0.888, 0.998) per unit GARD, p = 0.041). This association of GARD with OS also held when including only patients treated with primary definitive RT at standard of care doses (EQD2 69-71Gy) as shown in FIG. 12B (HR (95% CI) = 0.920 (0.857, 0.986) per unit GARD, p = 0.019). This suggests that GARD can stratify patients by predicted effect even when radiation dose is approximately uniform.

[0221] The cohort includes clinical variables for performance status (ECOG > 0), T stage (T4 vs TI-3), N stage (N2-N3 vs N0-N1), and smoking pack years (>10). Multivariable analysis was performed using these variables and GARD for statistical associations with OS. In the definitive primary RT cohort, GARD was the only variable statistically associated with OS (HR = 0.943 (0.891,0.999), p = 0.046). These results are summarized in Table 5, below.Attorney Docket No.: 146974.000006 PATENTTable 5 shows results from a multivariable analysis of definitive primary RT patients. GARD is associated with OS, censored at 60 months (p = 0.046).

[0222] Table 5.HR (95% CI) pGARD 0.943(0.891 0.999) 0.046T stage (T4) 1.992(0.711 5.576) 0.190N stage (N2-N3) 2.367(0.867 6.460) 0.093ECOG = 1 or 2 0.908(0.247 3.342) 0.884Pack-years (> 10) 2.117(0.756 5.929) 0.154

[0223] In addition, a Cox regression model was developed and evaluated including the previously known prognostic clinical variables (T stage, N stage, smoking and ECOG performance status), to determine whether a model including GARD improves overall model performance. As shown in FIG. 13, the Cox model including clinical variables achieves an AUC of 71.20 whereas GARD alone achieves a superior AUC (3 yr OS) of 78.26. Integrating GARD with the clinical variables improves the prognostic ability of the model with AUC 83.81. A 3-cluster model was evaluated which, by itself, achieves an AUC similar to the clinical variable model (AUC: 72.83). However, integration of the 3-cluster information into the GARD + clinical variable model does not improve the overall prognostic ability as measured by AUC 83.81. FIG.13 depicts AUC analysis showing GARD outperforms standard clinical variables. This model included all 191 definitive primary RT patients and analyzed outcome at 3 years. This lack of improvement may be related to the relationship between clusters and GARD values. See FIG.

[0224] Table 6 shows the time-dependent AUC and 95% CI at 3 years for the predictors individually as well as for the combined nomogram. Comparison of the 3 most significant predictors individually via ROC analysis showed the greatest AUC for GARD alone (78.26), while the model combining GARD, T stage, N stage, smoking and ECOG status produced the highest AUC (83.81); values are listed in Table 6. Of note, if RSI is compared here with the same cohort, a similar score to GARD is achieved of 77.68 (95% CI: 65.13 to 90.23) as the dose range in this cohort is narrow.

[0225] Table 6.3-yr AUC (95% CI)Clinical 71.20 (54.47, 89.93)Clusters 72.83 (59.01, 86.65)GARD 78.26 (65.14, 91.38)Clinical + GARD 83.81 (71.65, 95.97)Clinical + GARD + Clusters 83.81 (71.65, 95.97)Attorney Docket No.: 146974.000006 PATENT

[0226] GARD predicts that empiric dose de-escalation would result in inferior clinical outcome. Although HPV-positive patients have excellent prognosis, the interim analyses emphasized the importance of developing clinical tools to identify patient subsets with differential risk of clinical failure. It was hypothesized that GARD could identify a sub¬ population of HPV-positive patients at differential risk of failure that may explain the failure of unselected empiric dose de-escalation as tested. In addition, understanding the differential risks of failure can lead to a better clinical strategy for dose de-escalation in selected patients. To develop this, an exploratory discrete analysis based on an optimized cut-point analysis was performed. Minimizing the log-rank score at one discrete value reveals two groups with maximally different outcomes. See FIG. 15. This analysis revealed one cutpoint at GARD < 42 which optimally stratified patients as shown in FIG. 3D.

[0227] Patients that achieved the GARD high (GARD > 42) had a 3yr-OS of 100% (CI: 1-1) compared with 90% (CI: 0.85-0.96) for the GARD low group (GARD < 42). These differences are statistically significant with p = 0.0045.

[0228] One possible explanation is that empiric dose de-escalation results in a small number of patients falling from the GARD high cohort (GARD > 42) to the GARD low cohort (GARD < 42) leading to an inferior result for empiric dose de-escalation. To test this hypothesis, an in silico clinical trial was performed to evaluate GARD-based predictions of clinical outcome for empiric dose de-escalation to 60 Gy (with concurrent chemotherapy). GARD predicts that empiric (unselected) dose de-escalation would result in an inferior clinical outcome. The predicted 3 yr OS for patients modeled at 70 Gy is 94.6% compared with 92.7% for patients modeled at 60 Gy. See FIG. 16. Empiric, unselected dose de-intensification is predicted to increase the proportion of patients in the GARD low group while decreasing the proportion of patients in the GARD high group. The 70Gy in silico arm had an average of 126 and 74 patients in the low and high GARD groups, while the 60Gy in silico arm had 168 and 32 patients in those groups.

[0229] Next, it was determined whether GARD could be used to develop a clinical trial strategy that would predict equivalent outcome at 70 or 60 Gy. In one approach, GARD can identify patients that would remain at or above the GARD-high cutpoint (> 42) at 70 or 60 Gy. Based on simulations, approximately 16% of the HPV-positi ve trial populati on would be eligible for dose de-escalation in this scenario. This approach excludes GARD-low patients and patients that fall from GARD-high to GARD-low at 60 Gy. The predicted OS curve for this approach to de- escalation is shown in FIG. 17. The 36-month survival proportion is equivalent (94.6%) in both arms of this simulated trial.Attorney Docket No.: 146974.000006 PATENT

[0230] Overall, GARD predicts that uniform RT dose de-escalation in HPV-positive patients would result in an inferior clinical outcome compared to standard of care. FIG. 3D shows that GARD identifies HPV-positive patient subsets with differential risk of failure. An exploratory analysis identified one cut-point which group patients in two risk levels. Patients that achieve the lowest GARD (< 42) have a higher risk of failure (3-year OS = 90.5%). FIG. 19 shows in silico clinical trial designs. On the left side of FIG. 18, unselected RT dose de-escalation (cisplatin + 60 Gy vs cisplatin + 70 Gy) was selected. The RSI distribution of the cohort was used to generate GARD for 400 virtual patients randomized to either 60 or 70 Gy, and repeat this 100 times. The right side of FIG. 18 shows GARD-based de-escalation. In one example of a potential trial, a GARD-selected trial where only patients with GARD > 42 are eligible for randomization to de-intensification was simulated. FIG. 16 shows that simulation of unselected RT dose de-escalation (cisplatin + 60 Gy ) results in inferior OS compared to standard of care (cisplatin + 70 Gy ). The unselected in silico clinical trial predicts that patients treated with RT dose de-escalation experience a statistically significantly worse overall survival when compared to standard of care (3-yr OS 92.7% vs 94.6%, non-overlapping confidence intervals). FIG. 17 shows that selective de -intensification produces similar OS.

[0231] Another way to think about dose de-escalation is to ask the question: “can GARD identify a personalized target dose with the goal of maintaining current outcomes'!” This is fundamentally different than previous approaches which asked if GARD could be used to select patients for stratified de-escalation to standard dose levels. In this approach it is instead asked what GARD cutpoint would provide equipoise to current standard of care? Analyzing the cohort through this lens, it was found that a significantly lower GARD cutpoint equal or higher than 32 would provide outcomes in line with current standard of care. FIG. 19 shows the outcome of patients that achieved GARD 32 compared to unselected patients in the cohort. As shown, the patients that achieve a GARD of at least 32 have the same OS as the whole un-selected cohort, thus achieving equipoise with current SOC in unselected patients.

[0232] In a trial designed like this then, each patient would be assessed for their RSI, and then a physical radiation dose would be calculated such that they would achieve a prescribed GARD of at least 32. While exploratory and non-standard, this analysis of a genomic prescription paradigm offers a window into the future where dose is truly personalized - allowing exactly enough radiation to be delivered for tumor cure, minimizing toxicity. In FIG. 20 the minimum dose required for each patient in the cohort to achieve a GARD of at least 32 was calculated. Interestingly, the average dose needed aligns well with clinical intuition - approximately 60Gy - butAttorney Docket No.: 146974.000006 PATENTwith large heterogeneity across individual patients. Of note as well is the large number of patients (22.3 percent in this cohort, 39 / 175) who we predict require between 60 and 70 Gy, revealing which patients would have inferior outcomes when de-escalated to 60 Gy.

[0233] This also suggests that on average the toxicity (financial and clinical), of nearly 5 fractions / patient can be spared while maintaining similar outcomes. See Table 7, below. However, the potential toxicity reduction for each patient is variable with some patients predicted to only require 30 Gy while a small minority may require higher doses than standard. Of note, there is a peak in the distribution between 60 and 70Gy, meaning that as we reduce dose from 70-60Gy without genomic guidance, a significant portion of patients are underdosed, worsening outcomes. This stands in contrast to findings in non-small cell lung cancer, where the dose escalation from 60 to 74Gy (as in RTOG 0617) spanned a valley in the distribution, meaning that the escalation resulted in very few patients being benefited, while all received the increased toxicity.

[0234] Table 7.Personalized XRT SOCTotal predicted Fractions 5797 6685Fractions saved 888 -Mean fractions saved / patient 5.07 - 0235] FIG. 19 shows GARD targeted equipoise RT dose reduction. KM curves show that patients in the cohort that achieve a GARD of at least 32 achieve isocurative outcomes compared with the unselected cohort. Each patient can then have a prescription RT dose to match the target GARD (at least 32). Comparing this to SOC provides equivalent outcomes. FIG. 20 shows a histogram depicting the difference between the dose predicted to be required for each patient compared to the dose delivered. Patients on the right were potentially underdosed (39 / 175), offering opportunities to increase oncologic outcomes, and on the right were potentially overdosed, indicating opportunities to decrease toxicity. Table 7 calculates the difference between fractions delivered in SOC compared to the number predicted on a per patient basis reveals a large potential for toxicity reduction at the population level. This averages approximately 5 fractions (one week of radiotherapy) per patient, but with large heterogeneity.

[0236] With the aim of breaking the mold in radiation oncology trial design, two different GARD-based strategies are proposed to treatment optimization. The first prioritizes individual oncologic outcome for every patient, without considering concomitant toxicities at the population level. It was shown that personalizing RT dose to achieve GARD > 42 maximizes oncologic outcome in this dataset. However, only 16 percent of patients achieve GARD > 42 at 60 Gy, limiting this approach if the goal is to reduce toxicity. In the second approach, a risk of failure was defined that matches the risk of an unselected population treated at 70 Gy. ThisAttorney Docket No.: 146974.000006 PATENTallows a larger opportunity to reduce RT dose and toxicity risk. If the GARD model is based on these parameters, it is shown that that personalizing RT dose to achieve GARD of at least 32 achieves equipoise with current population level oncologic outcomes while reducing RT dose, and therefore toxicity, to the majority of patients.

[0237] The selection between a GARD value of about 41.9 (or about 42) and a GARD value of about 32 represents a choice between two distinct clinical treatment strategies. The higher GARD value (about 41.9 or 42) represents a strategy of maximizing individual patient cure probability, which may require higher radiation doses for some patients. The lower GARD value (about 32) represents a strategy of achieving population-level equipoise with current standard of care outcomes while enabling dose reduction for the majority of patients, thereby reducing overall toxicity burden. Both strategies are enabled by the GARD-based personalization approach, and the choice between them may be made by the clinician based on the specific clinical context and treatment goals.

[0238] In this approach, a larger proportion of patients become candidates for de-escalation (77.7%), while a small fraction was identified for potential escalation (22.3%). This strategy matches current clinical outcomes but delivers an average of 5 fewer fractions per patient and thus achieves the intended aim of equipoise at an average dose of 60 Gy. However, the difference is that the personalized dose for each patient is not 60 Gy, but instead, like most polygenic biological traits, a wide range (between 31-113 Gy). While most patients require lower doses, a small minority may need dose intensification or more a more effective radiosensitization strategy with concurrent chemotherapy.

[0239] In some embodiments, selecting a lower pre-determined GARD value, such as about 32, enables dose de-escalation for a substantial portion of the patient population. For example, in some embodiments, approximately 70% or more of oropharyngeal cancer patients may receive a lower radiation dose than the standard of care while maintaining equivalent clinical outcomes. This approach may reduce treatment-related toxicity at the population level while preserving the established cure rate. The selection of the lower GARD value thus represents a clinical strategy that prioritizes toxicity reduction across the patient population rather than maximizing individual cure probability.

[0240] Overall, GARD proposes a genomic-based strategy that achieves clinical equipoise while decreasing average dose by 5 fractions / patient. Thus, by accounting for biological heterogeneity, GARD-based RT prescription provides critical data that may improve the therapeutic ratio of RT; maximizing clinical outcome at the lowest possible toxicity risk. GARD predicts that while the majority of patients require lower doses than SOC, a small minority ofAttorney Docket No.: 146974.000006 PATENTresistant patients still require 70Gy or potentially a higher dose to maintain their excellent clinical outcome.

[0241] Definition of Minimal Clinical Target GARD Value and Recommended Effective Target GARD Value

[0242] The minimal clinical target GARD value of 32 refers to the lowest GARD threshold identified in oropharyngeal cancer patients that maintains OS rates equivalent to the current standard-of-care radiation therapy regimen. This threshold is based on retrospective and in silico analyses demonstrating that patients achieving a GARD of at least 32 exhibit comparable oncologic outcomes to an unselected patient cohort treated with standard radiation doses. The minimal GARD threshold serves as a baseline value to guide dose personalization while maintaining treatment efficacy.

[0243] The recommended effective target GARD value of 41.9 represents an optimized GARD threshold for predicting improved patient outcomes. This value was derived from statistical modeling of survival curves and discrete cut-point analyses, which demonstrated that a GARD of 41.9 or higher is associated with a higher probability of favorable OS. The selection of 41.9 as a recommended target reflects its ability to balance treatment efficacy and potential toxicity reduction, allowing for optimized radiation therapy planning.

[0244] In a GARD-based treatment planning approach, these thresholds guide radiation dosing by ensuring that: the minimum prescribed dose results in a GARD of at least 32, ensuring non-inferior oncologic outcomes; the optimal prescribed dose aims for a GARD of 41.9, maximizing therapeutic benefit while considering normal tissue toxicity constraints. By integrating these target values into radiation therapy planning, the system allows for precision dose adjustments tailored to each patient’s tumor radiosensitivity, minimizing unnecessary radiation exposure while maintaining or improving clinical efficacy.

[0245] In some embodiments, the selection of a pre-determined GARD value may be based on a clinical treatment priority. A first clinical treatment priority may be defined as maximizing the clinical effect of the radiation dosage, such as maximizing cancer cell kill or cure rate. In such embodiments, a higher GARD value, such as about 41.9 or about 42, may be selected. A second clinical treatment priority may be defined as minimizing off-target toxicity while maintaining an acceptable cure rate equivalent to current standard of care outcomes. In such embodiments, a lower GARD value, such as about 32, may be selected. The lower GARD value allows for dose de-escalation for a majority of patients while maintaining clinical outcomes equivalent to standard of care treatment. In some embodiments, the system or methodAttorney Docket No.: 146974.000006 PATENTmay receive input from a clinician indicating the desired clinical treatment priority, and may automatically select or recommend an appropriate GARD value based on the indicated priority.

[0246] Defining GARD Values Based on a Clinical Treatment Priority

[0247] In certain clinical scenarios, a clinician may wish to select a GARD value that corresponds to a prescribed dose that achieves a certain clinical outcome. For example, a clinician may wish to prioritize a maximum clinical effect of the dose, or in other words maximize a cell kill, over concerns about minimized off-target toxicity. In other scenarios, tire clinician may wish to prioritize minimizing the adverse off target toxicity effects while maintaining an established cure rate, in which case the GARD value corresponding to a lower dose can be selected. Accordingly, in certain embodiments, a method of calculating a radiation dosage (RxRSI) for the subject can be based at least in part on the RSI and a pre-determined GARD value, where the GARD value is selected as function of the treatment priority. In certain embodiments, tire personalized radiation therapy treatment plan for the subject can be based at least in part on the RxRSI and the treatment priority determined by the clinician.

[0248] In certain embodiments, actual treatment outcome data for the subject can be in the database (e.g., a reference population database), and the actual treatment outcome for each respective subject in the database can be compared to a predicted treatment outcome associated with the respective clinical treatment priority to determine whether the RxRSI dosage for the respecti ve patient met the desired treatment priori ty selected for that subject. If the comparison data shows that the selected RxRSI doses and associated GARD value for the selected clinical treatment priority do not align with one another, tire GARD value can be adjusted as needed until the selected RxRSI doses meet the desired treatment outcome for the selected priority. In other words, the GARD value can be adjusted so that the difference between the actual and predicted outcomes meets a predefined threshold, for example, a delta of zero, within 0.5%, within 1% or the like. The database can be continually updated as subjects are treated so that the GARD values align with the clinical treatment priorities to account for biological changes in the refence population (e.g., evolving treatment cocktails, changes in tumor response, changes in an average age of the reference population, or the like). Accordingly, clinicians can modulate the respective GARD value for each individuals treatment plan as needed based on the clinical treatment priority for that patient, and based on population data.

[0249] CLAUSES

[0250] Examples of the present disclosure can be implemented by any of the following numbered clauses:Attorney Docket No.: 146974.000006 PATENT

[0251] Clause 1. A computer-implemented method for personalizing radiation therapy based on clinical treatment priority, the method comprising: obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene¬ expression profiles for a reference population of tumors of a same or similar tumor type; obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor; determining a clinical treatment priority- selected from a first clinical treatment priority and a second clinical treatment priority; selecting a pre-determined genomic adjusted radiation dose (GARD) value based on the determined clinical treatment priority, wherein a first GARD value is associated with the first clinical treatment priority and a second GARD value, different from the first GARD value, is associated with the second clinical treatment priority; and calculating a radiation dosage (RxRSI) for the subject based at least in part on the RSI and the selected pre-determined GARD value.

[0252] Clause 2. The computer-implemented method of clause 1, wherein determining the clinical treatment priority further comprises obtaining the clinical treatment priority from user input.

[0253] Clause 3. The computer-implemented method of clause 1, wherein the first clinical treatment priority is defined as a first clinical goal of maximizing a clinical effect of the radiation dosage in the subject.

[0254] Clause 4. The computer-implemented method of clause 3, wherein the second clinical treatment priority is defined as a second clinical goal of minimizing off target toxicity of the radiation dosage in the subject while maintaining a predefined cure rate.

[0255] Clause 5. The computer-implemented method of clause 4, wherein the first GARD value is greater than the second GARD value.

[0256] Clause 6. The computer-implemented method of clause 4, further comprising: providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the clinical treatment priority.

[0257] Clause 7. The computer-implemented method of clause 5, further comprising: providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the determined clinical treatment priority; storing an actual treatment outcome for the subject in a database, the treatment outcome associated with the personalized radiation therapy treatment plan for the subject; comparing the actual treatment outcome for each respective subject in the database to a predicted treatment outcome associated with the respective clinical treatment priority for the respective subject to determine a treatment deltaAttorney Docket No.: 146974.000006 PATENTindicative of whether the respective RxRSI personalized radiation therapy treatment plan met the first clinical goal or the second clinical goal; and adjusting the first GARD value or the second GARD value as a function of the delta until the delta is within a predefined threshold.

[0258] Clause 8. A method of treating a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of the same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; selecting a clinical treatment priority; calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being selected as a function of the selected clinical treatment priority; providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the clinical treatment priority; storing treatment outcome data for the subject based on a treatment outcome of the personalized radiation therapy treatment plan for the subject to determine one or more of: an on-target efficacy for a reference population of subjects, an off-target toxicity effect for the reference population of subjects.

[0259] Clause 9. The method of clause 8, wherein the clinical treatment priority includes a first clinical treatment priority selected as a function of the on-target efficacy and a second clinical treatment priority selected as a function of the off-target toxicity effect.

[0260] Clause 10. The method of clause 9, wherein the first clinical treatment priority is a desired on-target efficacy and the second clinical treatment priority is a desired off-target toxicity effect.

[0261] Clause 11. The method of clause 10, wherein the desired on-target efficacy is a maximized cancer cell kill and the desired off-target toxicity effect is a minimum off-target toxicity effect that does not affect an overall known cure rate.

[0262] Clause 12. The method of clause 9, wherein the GARD value includes a first GARD value determined based on the first clinical treatment priority and a second GARD value determined based on the second clinical treatment priority.

[0263] Clause 13. The method of clause 12, further comprising: continually updating the database with the treatment outcome data; and adjusting the first GARD value and the second GARD value based on the updated database so that the first GARD value meets the first clinical treatment priority and the second GARD value meets the second clinical treatment priority.Attorney Docket No.: 146974.000006 PATENT

[0264] Clause 14. The method of clause 12, wherein the first GARD value is greater than the second GARD value.

[0265] Clause 15. A method of identifying a subject with oropharyngeal cancer for radiation therapy dose de-escalation, the method comprising: obtaining expression levels of one or more signature genes from a tumor sample of the subject; determining a radiation sensitivity index (RSI) of the subject's tumor sample based at least in part on the expression levels; calculating a genomic adjusted radiation dose (GARD) value for the subject based at least in part on the RSI and a proposed radiation dose; comparing the calculated GARD value to a pre-determined GARD threshold; identifying the subject as a candidate for dose de-escalation when the calculated GARD value meets or exceeds the pre-determined GARD threshold; and generating a radiation therapy treatment plan for the subject based at least in part on the identification, wherein the radiation therapy treatment plan specifies a de-escalated radiation dose for subjects identified as candidates for dose de-escalation.

[0266] Clause 16. A computer software configured to integrate with a radiation therapy¬ treatment planning system, the computer software being configured to: obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; assign a radiation sensitivity index (RSI) of the subject's tumor based at least in part on the expression levels of the one or more signature genes in the tumor; calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer; and provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0267] Clause 17. The computer software of clause 16, further being configured to: calculate the recommended RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

[0268] Clause 18. The computer software of clause 17, further being configured to: calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

[0269] Clause 19. The computer software of clause 16, further being configured to: receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the recommended RxRSI; calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans; penalize each radiation plan of the plurality of radiation plans basedAttorney Docket No.: 146974.000006 PATENTon the normal tissue toxicity of the respective radiation plan; and provide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

[0270] Clause 20. The computer software of clause 16, further being configured to: calculate the recommended RxRSI based in part on a predefined standard of care dose range.

[0271] Clause 21. The computer software of clause 16, further being configured to: calculate a proposed RxRSI for the subject based at least in part on tire pre-determined GARD value and the RSI; compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range; and recommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

[0272] Clause 22. The computer software of clause 16, further being configured to: apply a linear regression model to the expression levels of the one or more signature genes in the tumor; and assign the RSI based at least in part on the linear regression model.

[0273] Clause 23. A computer-implemented method for minimizing risk of radiation therapy comprising: obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor; calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer; and providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0274] Clause 24. The computer-implemented method of clause 23, further comprising: calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.

[0275] Clause 25. The computer-implemented mediod of clause 23, further comprising: calculating relative risk for potential RxRSI values; and selecting the RxRSI based at least in part on the relative risk.

[0276] Clause 26. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising: determining expression levels of one or more signature genesAttorney Docket No.: 146974.000006 PATENTthat characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

[0277] Clause 27. The method of clause 26, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.

[0278] Clause 28. The method of clause 26, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Ab 1 oncogene 1 (c-Ab 1); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (1RF1); and combinations thereof.

[0279] Clause 29. The method of clause 26, further comprising: determining a dose limiting structure of normal tissues based at least in part on the RxRSI.

[0280] Clause 30. A method of treating a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

[0281] Clause 31. The method of clause 30, further comprising: administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

[0282] Clause 32. The method of clause 30, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.

[0283] Clause 33. The method of clause 30, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator ofAttorney Docket No.: 146974.000006 PATENTtranscription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

[0284] Clause 34. A computer software configured to integrate with a radiation therapy treatment planning system, the computer software being configured to: obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously- measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; assign a radiation sensitivity index (RSI) of the subject's tumor based at least in part on the expression levels of the one or more signature genes in the tumor; calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer; and provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0285] Clause 35. The computer software of clause 34, further being configured to: calculate the recommended RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

[0286] Clause 36. The computer software of clause 35, further being configured to: calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

[0287] Clause 37. The computer software of clause 34, further being configured to: receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the recommended RxRSI; calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans; penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the respective radiation plan; and provide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

[0288] Clause 38. The computer software of clause 34, further being configured to: calculate the recommended RxRSI based in part on a predefined standard of care dose range.

[0289] Clause 39. The computer software of clause 34, further being configured to: calculate a proposed RxRSI for the subject based at least in part on the pre-determined GARD value and the RSI; compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when theAttorney Docket No.: 146974.000006 PATENTproposed RxRSI is within or below the predefined standard of care dose range; and recommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

[0290] Clause 40. The computer software of clause 34, further being configured to: apply a linear regression model to the expression levels of the one or more signature genes in the tumor; and assign the RSI based at least in part on the linear regression model.

[0291] Clause 41. A computer-implemented method for minimizing risk of radiation therapy comprising: obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; obtaining a radiation sensitivity index (RSI) of the subject’s tumor from the expression levels of the one or more signature genes in the tumor; calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre¬ determined GARD value being about 32 for oropharyngeal cancer; and providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0292] Clause 42. The computer-implemented method of clause 41, further comprising: calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.

[0293] Clause 43. The computer-implemented method of clause 41, further comprising: calculating relative risk for potential RxRSI values; and selecting the RxRSI based at least in part on the relative risk.

[0294] Clause 44. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar' tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer.

[0295] Clause 45. The method of clause 44, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.Attorney Docket No.: 146974.000006 PATENT

[0296] Clause 46. The method of clause 44, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

[0297] Clause 47. The method of clause 44, further comprising: determining a dose limiting structure of normal tissues based at least in part on the RxRSI.

[0298] Clause 48. A method of treating a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer.

[0299] Clause 49. The method of clause 48, further comprising: administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

[0300] Clause 50. The method of clause 48, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.

[0301] Clause 51. The method of clause 48, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

[0302] Clause 52. A computer software configured to integrate with a radiation therapy treatment planning system, the computer software being configured to: obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similarAttorney Docket No.: 146974.000006 PATENTtumor type; assign a radiation sensitivity index (RSI) of the subject's tumor based at least in part on the expression levels of the one or more signature genes in the tumor; calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre¬ determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer; and provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

[0303] Clause 53. The computer software of clause 52, further being configured to: calculate the recommended RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

[0304] Clause 54. The computer software of clause 53, further being configured to: calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

[0305] Clause 55. The computer software of clause 52, further being configured to: receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the recommended RxRSI; calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans; penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the respective radiation plan; and provide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

[0306] Clause 56. The computer software of clause 52, further being configured to: calculate the recommended RxRSI based in part on a predefined standard of care dose range.

[0307] Clause 57. The computer software of clause 52, further being configured to: calculate a proposed RxRSI for the subject based at least in part on the pre-determined GARD value and the RSI; compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range; and recommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

[0308] Clause 58. The computer software of clause 52, further being configured to: apply a linear regression model to the expression levels of the one or more signature genes in the tumor; and assign the RSI based at least in part on the linear regression model.

[0309] Clause 59. A computer-implemented method for minimizing risk of radiation therapy comprising: obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor orAttorney Docket No.: 146974.000006 PATENTselected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor; calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre -determined genomic adjusted radiation dose (GARD) value, the pre¬ determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer; and providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

[0310] Clause 60. The computer-implemented method of clause 59, further comprising: calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.

[0311] Clause 61. The computer-implemented method of clause 59, further comprising: calculating relative risk for potential RxRSI values; and selecting the RxRSI based at least in part on the relative risk.

[0312] Clause 62. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

[0313] Clause 63. The method of clause 62, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.

[0314] Clause 64. The method of clau se 62, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

[0315] Clause 65. The method of clause 62, further comprising: determining a dose limiting structure of normal tissues based at least in part on the RxRSI.Attorney Docket No.: 146974.000006 PATENT

[0316] Clause 66. A method of treating a subject, the method comprising: determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type; applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; and calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

[0317] Clause 67. The method of clause 66, further comprising: administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

[0318] Clause 68. The method of clause 66, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI of the subject based at least in part on the relative risk.

[0319] Clause 69. The method of clause 66, wherein determining the expression levels of the one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF 1); and combinations thereof.

[0320] Clause 70. A system for providing a radiation therapy treatment plan, the system comprising one or more processors and a memory storing instructions that, when executed by the one or more processors, cause the system to: obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject's tumor associated with oropharyngeal cancer or selected from a database of previously measured gene-expression profiles for a reference population of tumors associated with oropharyngeal cancer; assign a radiation sensitivity index (RSI) to the tumor associated with oropharyngeal cancer based at least in part on the expression levels of the one or more signature genes in the tumor; calculate a recommended personalized radiation dosage (RxRSI) for the subject with the tumor based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value; assign a minimal clinical target GARD value of 32; and assign a recommended effective target GARD value of 41.9.

[0321] Clause 71. The system of clause 70, wherein the instructions, when executed by the one or more processors, cause the system to: provide, based on the minimal clinical targetAttorney Docket No.: 146974.000006 PATENTGARD value and the recommended effective target GARD value, the recommended RxRSI as a radiation therapy dose for the radiation therapy treatment plan.

[0322] Clause 72. The system of clause 71, wherein providing the recommended RxRSI as the radiation therapy dose comprises displaying the recommended RxRSI on a display interface associated with the system.

[0323] Clause 73. The system of clause 71, wherein providing the recommended RxRSI as the radiation therapy dose comprises configuring a radiation therapy apparatus to deliver the recommended RxRSI to the subject.

Claims

Attorney Docket No.: 146974.000006 PATENTCLAIMSWhat Is Claimed Is:

1. A computer-implemented method for personalizing radiation therapy based on clinical treatment priority, the method comprising:obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor;determining a clinical treatment priority selected from a first clinical treatment priority and a second clinical treatment priority;selecting a pre-determined genomic adjusted radiation dose (GARD) value based on the determined clinical treatment priority, wherein a first GARD value is associated with the first clinical treatment priority and a second GARD value, different from the first GARD value, is associated with the second clinical treatment priority; andcalculating a radiation dosage (RxRSI) for the subject based at least in part on the RSI and the selected pre-determined GARD value.

2. The computer-implemented method of claim 1, wherein determining the clinical treatment priority further comprises obtaining the clinical treatment priority from user input.

3. The computer-implemented method of claim 1, wherein the first clinical treatment priority is defined as a first clinical goal of maximizing a clinical effect of the radiation dosage in the subject.

4. The computer-implemented method of claim 3, wherein the second clinical treatment priority is defined as a second clinical goal of minimizing off target toxicity of the radiation dosage in the subject while maintaining a predefined cure rate.

5. The computer-implemented method of claim 4, wherein the first GARD value is greater than the second GARD value.Attorney Docket No.: 146974.000006 PATENT6. The computer-implemented method of claim 4, further comprising: providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the determined clinical treatment priority.

7. The computer-implemented method of claim 6, further comprising:storing an actual treatment outcome for the subject in a database, the treatment outcome associated with the personalized radiation therapy treatment plan for the subject; comparing the actual treatment outcome for each respecti ve subject in the database to a predicted treatment outcome associated with the respective clinical treatment priority for the respective subject to determine a treatment delta indicative of whether the respective RxRSI personalized radiation therapy treatment plan met the first clinical goal or the second clinical goal; andadjusting the first GARD value or the second GARD value as a function of the delta until the delta is within a predefined threshold.

8. A method of treating a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of the same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample;selecting a clinical treatment priority;calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being selected as a function of the selected clinical treatment priority;providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI and the clinical treatment priority;storing treatment outcome data for the subject based on a treatment outcome of the personalized radiation therapy treatment plan for the subject to determine one or more of: anAttorney Docket No.: 146974.000006 PATENTon-target efficacy for the reference population, an off-target toxicity effect for the reference population.

9. The method of claim 8, wherein the clinical treatment priority includes a first clinical treatment priority selected as a function of the on-target efficacy and a second clinical treatment priority is selected as a function of the off-target toxicity effect.

10. The method of claim 9, wherein the first clinical treatment priority is a desired on- target efficacy and the second clinical treatment priority is a desired off-target toxicity effect.

11. The method of claim 10, wherein the desired on-target efficacy is a maximized cancer cell kill and the desired off-target toxicity effect is a minimum off-target toxicity effect that does not affect an overall known cure rate.

12. The method of claim 11, wherein the GARD value includes a first GARD value determined based on the first clinical treatment priority and a second GARD value determined based on the second clinical treatment priority.

13. The method of claim 12, further comprising:continually updating the database with the treatment outcome data; and adjusting the first GARD value and the second GARD values based on the updated database so that the first GARD value meets the first clinical treatment priority and the second GARD value meets the second clinical treatment priority.

14. The method of any of claims 8-13, wherein the first GARD value is greater than the second GARD value.

15. A method of identifying a subject with oropharyngeal cancer for radiation therapy dose de-escalation, the method comprising:obtaining expression levels of one or more signature genes from a tumor sample of the subject;determining a radiation sensitivity index (RSI) of the subject’s tumor sample based at least in part on the expression levels;calculating a genomic adjusted radiation dose (GARD) value for the subject based at least in part on the RSI and a proposed radiation dose;Attorney Docket No.: 146974.000006 PATENTcomparing the calculated GARD value to a pre-determined GARD threshold; identifying the subject as a candidate for dose de-escalation when the calculated GARD value meets or exceeds the pre-determined GARD threshold; andgenerating a radiation therapy treatment plan for the subject based at least in part on the identification, wherein the radiation therapy treatment plan specifies a de-escalated radiation dose for subjects identified as candidates for dose de-escalation.

16. A computer software configured to integrate with a radiation therapy treatment planning system, the computer software being configured to:obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;assign a radiation sensitivity index (RSI) of the subject’s tumor based at least in part on the expression levels of the one or more signature genes in the tumor;calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre -determined GARD value being about 41.9 or about 32 for oropharyngeal cancer; andprovide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

17. The computer software of claim 16, further being configured to:calculate the RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

18. The computer software of claim 16, further being configured to:calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

19. The computer software of claim 16, further being configured to:receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the RxRSI;Attorney Docket No.: 146974.000006 PATENTcalculate normal tissue toxicity for each radiation plan of the plurality of radiation plans;penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the radiation treatment plan; andprovide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

20. The computer software of claim 16, further being configured to:calculate the recommended RxRSI based in part on a predefined standard of care dose range.

21. The computer software of claim 16, further being configured to:calculate a proposed RxRSI based for tlie subject based at least in part on the predetermined GARD value and the RSI;compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range; andrecommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

22. The computer software of claim 16, further being configured to:apply a linear regression model to the expression levels of the one or more signature genes in the tumor; andassign the RSI based at least in part on the linear regression model.

23. A computer-implemented method for minimizing risk of radiation therapy comprising:obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor;Attorney Docket No.: 146974.000006 PATENTcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer: and providing a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

24. The computer-implemented method of claim 23, further comprising:calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.

25. The computer-implemented method of claim 23, further comprising:calculating relative risk for potential RxRSI values; and selecting the RxRSI based at least in part on the relative risk.

26. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample: andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

27. The method of claim 26, further comprising: calculating relative risk for the RxRSI; and selecting the RxRSI value of the subject based at least in part on the relative risk.

28. The method of claim 26, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1Attorney Docket No.: 146974.000006 PATENT(STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

29. The method of claim 26, further comprising:determining a dose limiting structure of the normal tissues based at least in the RxRSI.

30. A method of treating a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample: andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 41.9 or about 32 for oropharyngeal cancer.

31. The method of claim 30, further comprising:administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

32. The method of claim 30, further comprising:calculating relative risk for the RxRSI; andselecting the RxRSI value of the subject based at least in part on the relative risk.

33. The method of claim 30, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21Attorney Docket No.: 146974.000006 PATENTactivated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

34. A computer software configured to integrate with a radiation therapy treatment planning system, the computer software being configured to:obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;assign a radiation sensitivity index (RSI) of the subject’s tumor based at least in part on the expression levels of the one or more signature genes in the tumor;calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer; and provide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

35. The computer software of claim 34, further being configured to:calculate the RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

36. The computer software of claim 34, further being configured to:calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

37. The computer software of claim 34, further being configured to:receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the RxRSI;calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans;penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the radiation treatment plan; andprovide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.Attorney Docket No.: 146974.000006 PATENT38. The computer software of claim 34, further being configured to:calculate the recommended RxRSI based in part on a predefined standard of care dose range.

39. The computer software of claim 34, further being configured to:calculate a proposed RxRSI based for the subject based at least in part on the pre¬ determined GARD value and the RSI;compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range; andrecommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

40. The computer software of claim 34, further being configured to:apply a linear regression model to the expression levels of the one or more signature genes in the tumor; andassign the RSI based at least in part on the linear regression model.

41. A computer-implemented method for minimizing risk of radiation therapy comprising:obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor;calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer; andproviding a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.Attorney Docket No.: 146974.000006 PATENT42. The computer-implemented method of claim 41, further comprising:calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.

43. The computer-implemented method of claim 41, further comprising:calculating relative risk for potential RxRSI values; andselecting the RxRSI based at least in part on the relative risk.

44. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer.

45. The method of claim 44, further comprising:calculating relative risk for the RxRSI; andselecting the RxRSI value of the subject based at least in part on the relative risk.

46. The method of claim 44, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRFl); and combinations thereof.

47. The method of claim 44, further comprising:Attorney Docket No.: 146974.000006 PATENTdetermining a dose limiting structure of the normal tissues based at least in the RxRSI.

48. A method of treating a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being about 32 for oropharyngeal cancer.

49. The method of claim 48, further comprising:administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

50. The method of claim 48, further comprising:calculating relative risk for the RxRSI; andselecting the RxRSI value of the subject based at least in part on the relative risk.

51. The method of claim 48, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

52. A computer software configured to integrate with a radiation therapy treatment planning system, the computer software being configured to:obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from aAttorney Docket No.: 146974.000006 PATENTdatabase of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;assign a radiation sensitivity index (RSI) of the subject’s tumor based at least in part on the expression levels of the one or more signature genes in the tumor;calculate a recommended personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer; andprovide, to the radiation therapy treatment planning system, the recommended RxRSI as a radiation therapy dose for a radiation plan.

53. The computer software of claim 52, further being configured to:calculate the RxRSI based in part on normal tissue toxicity for a treatment plan using the recommended RxRSI.

54. The computer software of claim 52, further being configured to:calculate the normal tissue toxicity based at least in part on risks to a plurality of tissue sites.

55. The computer software of claim 52, further being configured to:receive, from the radiation therapy treatment planning system, a plurality of radiation plans each using the RxRSI;calculate normal tissue toxicity for each radiation plan of the plurality of radiation plans;penalize each radiation plan of the plurality of radiation plans based on the normal tissue toxicity of the radiation treatment plan; andprovide, to the radiation therapy treatment planning system, at least one recommended radiation plan that is least penalized of the plurality of radiation plans.

56. The computer software of claim 52, further being configured to:calculate the recommended RxRSI based in part on a predefined standard of care dose range.

57. The computer software of claim 52, further being configured to:Attorney Docket No.: 146974.000006 PATENTcalculate a proposed RxRSI based for the subject based at least in part on the pre¬ determined GARD value and the RSI;compare the proposed RxRSI to a predefined standard of care dose range; assign the recommended RxRSI a value within the predefined standard of care dose range when the proposed RxRSI is within or below the predefined standard of care dose range; andrecommend consideration of the subject for clinical trial if the proposed RxRSI is above the predefined standard of care dose range.

58. The computer software of claim 52, further being configured to:apply a linear regression model to the expression levels of the one or more signature genes in the tumor; andassign the RSI based at least in part on the linear regression model.

59. A computer-implemented method for minimizing risk of radiation therapy comprising:obtaining or retrieving expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;obtaining a radiation sensitivity index (RSI) of the subject's tumor from the expression levels of the one or more signature genes in the tumor;calculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer; andproviding a personalized radiation therapy treatment plan for the subject based at least in part on the RxRSI.

60. The computer-implemented method of claim 59, further comprising:calculating normal tissue toxicity of the personalized radiation dosage; and providing the personalized radiation therapy treatment plan based at least in part on the normal tissue toxicity.Attorney Docket No.: 146974.000006 PATENT61. The computer-implemented method of claim 59, further comprising:calculating relative risk for potential RxRSI values; andselecting the RxRSI based at least in part on the relative risk.

62. A method of calculating a personalized radiation therapy dosage for a subject, the method comprising:determining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

63. The method of claim 62, further comprising:calculating relative risk for the RxRSI; andselecting the RxRSI value of the subject based at least in part on the relative risk.

64. The method of claim 62, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

65. The method of claim 62, further comprising:determining a dose limiting structure of the normal tissues based at least in the RxRSI.

66. A method of treating a subject, the method comprising:Attorney Docket No.: 146974.000006 PATENTdetermining expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor sample or selected from a database of previously measured gene-expression profiles for a reference population of tumors of a same or similar tumor type;applying a linear regression model to the expression levels and assigning a radiation sensitivity index (RSI) to the subject's tumor sample; andcalculating a personalized radiation dosage (RxRSI) for the subject based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value, the pre-determined GARD value being from about 32 to about 41.9 for oropharyngeal cancer.

67. The method of claim 66, further comprising:administering the calculated personalized radiation dosage (RxRSI) to the subject as a treatment for oropharyngeal cancer.

68. The method of claim 66, further comprising:calculating relative risk for the RxRSI; andselecting the RxRSI value of the subject based at least in part on the relative risk.

69. The method of claim 66, wherein determining the expression levels of one or more signature genes comprises determining the expression levels of genes selected from androgen receptor (AR); jun oncogene (c-Jun); signal transducer and activator of transcription 1 (STAT1); protein kinase C, beta (PKC); V-rel reticuloendotheliosis viral oncogene homolog A (RELA or p65); c-Abl oncogene 1 (c-Abl); small ubiquitin-like modifier 1 (SUMO1); p21 activated kinase-2 (PAK2); histone deacetylase 1 (HDAC1); interferon regulatory factor 1 (IRF1); and combinations thereof.

70. A system for providing a radiation therapy treatment plan, the system comprising one or more processors and a memory storing instructions that, when executed by the one or more processors, causes the system to:obtain or retrieve expression levels of one or more signature genes that characterize a tumor, wherein the expression levels are measured in a subject’s tumor associated with oropharyngeal cancer or selected from a database of previously measured gene-expression profiles for a reference population of tumors associated with oropharyngeal cancer;Attorney Docket No.: 146974.000006 PATENTassign a radiation sensitivity index (RSI) to the tumor associated with oropharyngeal cancer based at least in part on the expression levels of the one or more signature genes in the tumor;calculate a recommended personalized radiation dosage (RxRSI) for a subject with the tumor based at least in part on the RSI and a pre-determined genomic adjusted radiation dose (GARD) value;assign a minimal clinical target GARD value of 32; andassign a recommended effective target GARD value of 41.9.

71. The system of Claim 70, wherein the instructions, when executed by the one or more processors, cause the system to:provide, based on the minimal clinical target GARD value and the recommended effective target GARD, the recommended RxRSI as a radiation therapy dose for the radiation therapy treatment plan.

72. The system of Claim 71, wherein providing the recommended RxRSI as the radiation therapy dose comprises displaying the recommended RxRSI on a display interface associated with the system.

73. The system of Claim 71, wherein providing the recommended RxRSI as the radiation therapy dose comprises configuring a radiation therapy apparatus to deliver the recommended RxRSI to the subject.

Images