METHODS FOR DETERMINING TREATMENT FOR CANCER PATIENTS
Methods for determining treatments for cancer patients are disclosed.
The present invention generally relates to methods of treating patients for cancer.
BACKGROUND OF THE INVENTIONDevelopment of new anticancer therapeutics, such as immuno-oncology therapeutics (TO), drugs targeting DNA damage repair systems, and the like, requires extensive investments of time and resources to identify potentially effective candidate drugs. However, before such candidates can be added to the existing armamentarium of approved anticancer therapies, they must be proven to be both safe and effective in clinical trials. The initial step in this process is to conduct phase I clinical trials, which are designed to focus on drug safety.
These phase I clinical trials recruit patients who may potentially benefit from taking part in the trial, oftentimes because standard of care treatments for these patients have failed. In many clinical trials, irrespective of the type of treatment being tested or its intended indication, a subset of the participants is unable to complete the study due to a variety of reasons. For example, some patients may enter clinical trials with significantly advanced underlying disease or may have secondary or other unrelated pathological conditions to their cancer.
In the context of IO phase I clinical trials, such as PD-1 or PD-L1 (PDx) studies as monotherapies, nearly 20% of patients recruited to participate die within 2-3 months of beginning the study, even though ideally most trials of this type last much longer than 2-3 months. Therefore, to participate in an IO phase I clinical trial, patients should have a life expectancy of greater than 24 weeks and not be at risk of early mortality (EM; defined as a life expectancy of less than 24 weeks). An EM phenomenon has been observed in many randomized clinical trials comparing ICIs with active comparator arms in advanced or metastatic cancer patients, even with overall benefit ultimately favoring ICI therapy (Champiat, et al. 2018). While the precise etiology of this phenomenon is not clearly established, it is characterized by what seems to be disproportionately higher mortality in the early treatment period where survival is favored in the control arm, followed by subsequent benefit in overall survival (OS) favoring the ICI treatment arm. This is often reflected in the clinical data by the “crossing of the Kaplan-Meier curves” suggesting a subpopulation of patients at a higher risk of EM whose advanced rate of tumour growth may require the cytotoxic tumour-debulking effect of chemotherapy.
Accordingly, it is of great therapeutic interest to better predict the risk of early mortality for a given patient to better inform the appropriate treatment for their individual clinical state. An unexpected rapid disease progression or exacerbation of another underlying pathological condition in a significantly large number of patients participating in a phase I clinical trial can diminish the ability to detect early efficacy signals of the therapeutic being tested, delay dose escalation due to premature patient dropout, confound analysis of predictive biomarkers of efficacy, and dilute efficacy versus comparators in randomized trials. This can also prevent patients from getting potentially life-extending therapeutic intervention outside of a phase I clinical trial. If such patients were to receive therapeutic intervention that corrects or alleviates the underlying condition, they may then be able to fully benefit from trial participation. Therefore, improved prognostic methodologies for predicting EM in patients for anticancer therapeutic clinical trials are needed.
Immune checkpoint inhibitors (ICIs) are profoundly changing the treatment of many types of cancer, including melanoma, non-small cell lung cancer (NSCLC), renal cell carcinoma (RCC), squamous cell carcinoma of the head and neck (SCCHN), urothelial carcinoma (UC), and Hodgkin's lymphoma, and have been associated with long-lasting tumour responses. To aid in this objective, herein is presented a novel prognostic score, Immune Immediacy Index (3i score), that uses pre-treatment measurements of routinely collected blood-based factors to predict a patient's risk of early mortality, and to optimize the benefit:risk profile for treatment of patients with ICI.
SUMMARY OF THE INVENTIONAs described below, in a first aspect, the present disclosure provides a method of treating a patient having cancer, comprising: a) determining concentrations of multiple clinicopathological markers of the patient; b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient is likely to die within a period of 12 weeks; and d) administering an anticancer therapeutic to the patient if the prediction indicates that the patient is not likely to die within the period of 12 weeks.
In one embodiment of the first aspect, the multiple clinicopathological markers comprise one or more of age, albumin (g/L), alanine phosphatases (U/L), alanine amino transferase (U/L), aspartate amino transferase (U/L), basophils (103/μL), basophils/Leukocytes (%), bilirubin (μmol/L), Body Mass Index, calcium (mmol/L), chloride (mmol/L), eosinophils (103/μL), eosinophils/leukocytes (%), gamma-glutamyl transferase (U/L), glucose (mmol/L), hematocrit (%), hemoglobin (g/L), potassium (mmol/L), lactate dehydrogenase (U/L), liver metastasis (yes/no), lymphocytes (103/μL), lymphocytes/leukocytes (%), magnesium (mmol/L), monocytes (103/μL), monocytes/leukocytes (%), neutrophils (103/μL), neutrophils/leukocytes (%), neutrophils/lymphocytes (%), platelets (103/μL), protein (g/L), gender, sodium (mmol/L), thyrotropin (mU/L), tumor size, uric acid (mmol/L), and leukocytes (103/μL).
In another embodiment of the first aspect, the multiple clinicopathological markers comprise neutrophils/leukocytes (%), neutrophils (103/μL), albumin (g/L), lactate dehydrogenase (U/L), aspartate amino transferase (U/L), and/or gamma-glutamyl transferase (U/L).
In one embodiment of the first aspect, the method further comprises taking a sample from the patient.
In one embodiment of the first aspect, the sample comprises a tissue biopsy, cerebrospinal fluid, lymph, whole blood, serum, a blood cell, urine, sweat, tears, saliva, and/or feces.
In one embodiment of the first aspect, a concentration of one or more of the multiple clinicopathological markers is determined from the sample taken from the patient.
In one embodiment of the first aspect, the anticancer therapeutic comprises cisplatin, gemcitabine, methotrexate, vinblastine, doxorubicin, cisplatin (MVAC), carboplatin, a taxane, temozolomide, dacarbazine, vinflunine, docetaxel, paclitaxel, nab-paclitaxel, Vemurafenib, Erlotinib, Afatinib, Cetuximab, Bevacizumab, Gefitinib, and/or Pemetrexed.
In one embodiment of the first aspect, the anticancer therapeutic comprises an immune checkpoint inhibitor.
In one embodiment of the first aspect, the immune checkpoint inhibitor comprises an anti-CTLA-4 antibody, an anti-PD-1 antibody, and/or an anti-PD-L1 antibody.
In one embodiment of the first aspect, the anti-CTLA-4 antibody is tremelimumab or ipilimumab.
In one embodiment of the first aspect, the anti-PD-1 antibody is REGN2810, SHR1210, IBI308, PDR001, nivolumab, pembrolizumab, Anti-PD-1, BGB-A317, BCD-100, or JS001.
In one embodiment of the first aspect, the anti-PD-L1 antibody comprises durvalumab, avelumab, atezolizumab, or KNO35.
In a further embodiment of the first aspect, the method further includes determining a cancer type of the patient and also providing, to the gradient boosting machine learning model, the determined cancer type.
In a second aspect, the present disclosure provides a method of treating a patient having cancer, comprising: a) determining concentrations of multiple clinicopathological markers of the patient; b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has poor immune fitness; and d) administering an anticancer therapeutic to the patient if the prediction indicates that the patient does not have poor immune fitness.
In one embodiment of the second aspect, the method further includes e) reducing in the patient at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of tumor metastasis.
In a third aspect, the present disclosure provides a method of treating a patient having cancer, comprising: a) determining concentrations of multiple clinicopathological markers of the patient; b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has poor immune fitness; d) administering a therapeutic to the patient to boost the patient's immune fitness if the prediction indicates that the patient has poor immune fitness; e) determining second concentrations of multiple clinicopathological markers of the patient; f) providing, to a gradient boosting machine learning model, the determined second concentrations of the multiple clinicopathological markers; g) receiving, from the gradient boosting machine learning model, a second prediction of whether the patient has poor immune fitness; and h) administering an anticancer therapeutic to the patient if the prediction indicates that the patient does not have poor immune fitness.
In a fourth aspect, the present disclosure provides a method of conducting a clinical trial for an anticancer therapeutic, comprising: a) identifying a candidate participant for a clinical trial, wherein the candidate participant has cancer; b) determining concentrations of multiple clinicopathological markers of the patient; c) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; d) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has a likelihood of death within a period of 12 weeks; and e) including the patient in the clinical trial for the anticancer therapeutic if the prediction indicates that the patient does not have a likelihood of death within the period of 12 weeks.
In one embodiment of the fourth aspect, the anticancer therapeutic is an immune checkpoint inhibitor.
In a fifth aspect, the present disclosure provides a method of predicting the effectiveness of treating a cancer patient with an immune checkpoint inhibitor, comprising: a) determining concentrations of multiple clinicopathological markers of the patient; b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient lacks immune fitness, wherein a prediction indicating that the patient does not lack immune fitness predicts that the patient may be more effectively treated with an immune checkpoint inhibitor (ICI) than a patient predicted to lack immune fitness.
In a sixth aspect, the present disclosure provides a method of determining a treatment regimen for a cancer patient, comprising: a) determining concentrations of multiple clinicopathological markers of the patient; b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers; c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient lacks immune fitness, wherein a prediction indicating that the patient does not lack immune fitness suggests a treatment regimen including an immune checkpoint inhibitor (ICI), and wherein a prediction indicating the patient lacks immune fitness suggests a treatment regimen without an immune checkpoint inhibitor (ICI).
In a seventh aspect, a computing system includes computer memory configured to store model training data, wherein the model training data involves a first set of cancer patients that underwent cancer treatments, wherein the model training data associates (i) results from laboratory tests conducted on the first set of cancer patients and tumor types of the first set of cancer patients with (ii) whether individuals from the first set of cancer patients died within a threshold number of weeks from initiation of the cancer treatments. The computing system also includes one or more processors in communication with the computer memory and configured to execute program instructions to: (i) train, based on the model training data, a sequential series of decision trees until a stopping condition is reached, wherein each subsequent decision tree of the sequential series of decision trees is constructed based on residual values of its preceding decision tree of the sequential series of decision trees; and (ii) generate a gradient boosting machine learning model as an additive function of the sequential series of decision trees, wherein the gradient boosting machine learning model predicts whether a further cancer patient dies within the threshold number of weeks based on the additive function applied to results from the laboratory tests as conducted on the further cancer patient and a tumor type of the further cancer patient.
In an eighth aspect, a computer implemented method involves obtaining model training data, wherein the model training data involves a first set of cancer patients that underwent cancer treatments, wherein the model training data associates (i) results from laboratory tests conducted on the first set of cancer patients and tumor types of the first set of cancer patients with (ii) whether individuals from the first set of cancer patients died within a threshold number of weeks from initiation of the cancer treatments. The computer implemented method also involves training, based on the model training data, a sequential series of decision trees until a stopping condition is reached, and wherein each subsequent decision tree of the sequential series of decision trees is constructed based on residual values of its preceding decision tree of the sequential series of decision trees. The computer implemented method also involves generating a gradient boosting machine learning model as an additive function of the sequential series of decision trees, wherein the gradient boosting machine learning model predicts whether a further cancer patient dies within the threshold number of weeks based on the additive function applied to results from the laboratory tests as conducted on the further cancer patient and a tumor type of the further cancer patient.
In a ninth aspect, an article of manufacture includes a non-transitory computer-readable medium, having stored thereon program instructions that, upon execution by a computing device, cause the computing device to perform the operations of the eighth aspect.
In a tenth aspect, a computing system involves computer memory configured to store a gradient boosting machine learning model, wherein the gradient boosting machine learning model was trained on model training data, wherein the model training data involves a first set of cancer patients that underwent cancer treatments, wherein the model training data associates (i) results from laboratory tests conducted on the first set of cancer patients and tumor types of the first set of cancer patients with (ii) whether individuals from the first set of cancer patients died within a threshold number of weeks from initiation of the cancer treatments, wherein training the gradient boosting machine learning model involved: (i) training a sequential series of decision trees until a stopping condition was reached, wherein each subsequent decision tree of the sequential series of decision trees was constructed based on residual values of its preceding decision tree of the sequential series of decision trees, and (ii) determining an additive function of the sequential series of decision trees. The computing system also involves one or more processors in communication with the computer memory and configured to execute program instructions to: (i) obtain results from the laboratory tests as conducted on a further cancer patient; (ii) obtain a tumor type of the further cancer patient; (iii) apply the gradient boosting machine learning model to results from the laboratory tests as conducted on the further cancer patient and the tumor type of the further cancer patient; and (iv) receive, from the gradient boosting machine learning model, a prediction of whether the further cancer patient dies within the threshold number of weeks.
In an eleventh aspect, a computer-implemented method involves obtaining a gradient boosting machine learning model, wherein the gradient boosting machine learning model was trained on model training data, wherein the model training data involves a first set of cancer patients that underwent cancer treatments, wherein the model training data associates (i) results from laboratory tests conducted on the first set of cancer patients and tumor types of the first set of cancer patients with (ii) whether individuals from the first set of cancer patients died within a threshold number of weeks from initiation of the cancer treatments, wherein training the gradient boosting machine learning model involved: (i) training a sequential series of decision trees until a stopping condition was reached, wherein each subsequent decision tree of the sequential series of decision trees was constructed based on residual values of its preceding decision tree of the sequential series of decision trees, and (ii) determining an additive function of the sequential series of decision trees. The computer-implemented method also involves obtaining results from the laboratory tests as conducted on a further cancer patient. The computer-implemented method also involves obtaining a tumor type of the further cancer patient. The computer-implemented method also involves applying the gradient boosting machine learning model to results from the laboratory tests as conducted on a further cancer patient and the tumor type of the further cancer patient. The computer-implemented method also involves receiving, from the gradient boosting machine learning model, a prediction of whether the further cancer patient dies within the threshold number of weeks.
In a twelfth aspect, an article of manufacture includes a non-transitory computer-readable medium, having stored thereon program instructions that, upon execution by a computing device, cause the computing device to perform the operations of the eleventh aspect.
Other features and advantages of the invention will be apparent from the detailed description, and the claims.
Unless defined otherwise, all technical and scientific terms used herein have the meaning commonly understood by a person skilled in the art to which this invention belongs. The following references provide one of skill with a general definition of many of the terms used in this invention: Singleton, et al., Dictionary of Microbiology and Molecular Biology (2nd ed. 1994); The Cambridge Dictionary of Science and Technology (Walker ed., 1988); The Glossary of Genetics, 5th Ed., R. Rieger, et al. (eds.), Springer Verlag (1991); and Hale & Marham, The Harper Collins Dictionary of Biology (1991). As used herein, the following terms have the meanings ascribed to them below, unless specified otherwise.
By “anti-PD-L1 antibody” is meant an antibody that selectively binds a PD-L1 polypeptide. Exemplary anti-PD-L1 antibodies are described for example at U.S. Pat. Nos. 8,779,108; 9,493,565; and 10,400,039 which are herein incorporated by reference, which is herein incorporated by reference. Durvalumab (MEDI4736), or “Durva,” is an exemplary anti-PD-L1 antibody that is suitable for the methods described herein.
By “an anti-CTLA4 antibody” is meant an antibody that selectively binds a CTLA4 polypeptide. Exemplary anti-CTLA4 antibodies are described for example at U.S. Pat. Nos. 6,682,736; 7,109,003; 7,123,281; 7,411,057; 7,824,679; 8,143,379; 7,807,797; and 8,491,895 (Tremelimumab is 11.2.1, therein), which are herein incorporated by reference. Tremelimumab, or “Treme,” is an exemplary anti-CTLA4 antibody.
The term “biomarker,” “marker,” or “clinicopathological marker” (which can be used interchangeably) as used herein generally refers to a protein, nucleic acid molecule, clinical indicator, or other analyte that is associated with a disease. In one embodiment, a clinicopathological marker can be differentially expressed (or present) in a biological sample obtained from a subject having a disease (e.g., bladder cancer) relative to the concentration present in a control sample or reference.
In another embodiment, a clinicopathological marker is an indicator of extent of a disease. For example, when a concentration of a clinicopathological marker in a patient having a disease (e.g., cancer) is elevated compared to control (for example, a subject without cancer), then the elevated concentration can be indicative of disease progression in the patient. Similarly, when a concentration of a clinicopathological marker in a patient having a disease (e.g., cancer) is elevated compared to an earlier measurement of the marker in the patient herself, then the elevated concentration can be indicative of disease progression in the patient. Further, in some embodiments, a concentration of a clinicopathological marker can be indicative of a patient's relative health (e.g., body mass index) and/or the patient's general physiological resilience (e.g., age).
In some embodiments, concentrations of one or more clinicopathological markers can be indicative of a patient's immune fitness, for example, the relative ability of the patient's immune system to combat a particular disease by itself or the relative ability of the patient's immune system to be augmented or modified to combat a particular disease, such as cancer, by treatment with an therapeutic agent such as an immune checkpoint inhibitor (ICI).
In this disclosure, “comprises,” “comprising,” “containing” and “having” and the like can have the meaning ascribed to them in U.S. Patent law and can mean “includes,” “including,” and the like; “consisting essentially of” or “consists essentially” likewise has the meaning ascribed in U.S. Patent law and the term is open-ended, allowing for the presence of more than that which is recited so long as basic or novel characteristics of that which is recited is not changed by the presence of more than that which is recited, but excludes prior art embodiments.
As used herein, the terms “determining”, “assessing”, “assaying”, “measuring” and “detecting”, and “identifying” refer to both quantitative and qualitative determinations, and as such, the term “determining” is used interchangeably herein with “assaying,” “measuring,” and the like. Where a quantitative determination is intended, the phrase “determining an amount” of an analyte, substance, protein, and the like is used. Where a qualitative and/or quantitative determination is intended, the phrase “determining a concentration” of an analyte or “detecting” an analyte is used.
By “disease” is meant any condition or disorder that damages, interferes with or dysregulates the normal function of a cell, tissue, or organ. In a disease such as cancer (e.g., bladder cancer) the normal function of a cell tissue or organ is subverted to enable immune evasion and/or escape.
By “responsive” in the context of therapy is meant susceptible to treatment. By “subject” is meant a mammal, including, but not limited to, a human, such as a human patient, a non-human primate, or a non-human mammal, such as a bovine, equine, canine, ovine, or feline animal.
As used herein, the terms “treat,” treating,” “treatment,” and the like refer to reducing, diminishing, lessening, alleviating, abrogating, or ameliorating a disorder and/or symptoms associated therewith. It will be appreciated that, although not precluded, treating a disorder or condition does not require that the disorder, condition or symptoms associated therewith be completely eliminated.
Unless specifically stated or obvious from its context, the term “or” as used herein is understood to be inclusive. Unless specifically stated or obvious from context, the terms “a”, “an”, and “the” as used herein are understood to be singular or plural. Similarly, a particular term when expressed in the singular form also contemplates the same term expressed in plural, and vice versa. For example, the term “drug” also contemplates “drugs” and vice versa.
Unless specifically stated or obvious from context, as used herein, the term “about” is understood as within a range of normal tolerance in the art, for example within 2 standard deviations of the mean. The term “about” is understood to refer to within 5%, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%, 0.5%, 0.1%, 0.05%, or 0.01% of the stated value. Unless otherwise clear from context, all numerical values provided herein are modified by the term about.
The recitation of a listing of chemical groups in any definition of a variable herein includes definitions of that variable as any single group or combination of listed groups. The recitation of an embodiment for a variable or aspect herein includes that embodiment as any single embodiment or in combination with any other embodiments or portions thereof.
Early MortalityAs described herein, the present invention features, in one embodiment, a method of predicting whether a cancer patient is a fast progressor or is at risk for early mortality (EM; a patient whose overall survival (OS) is 24 weeks or less) by taking a blood sample or other samples and physiological measurements as described herein elsewhere, measuring concentrations of clinicopathological markers, and generating a composite score of the marker concentrations, wherein the clinicopathological markers can be, for example, neutrophils, hemoglobin, aspartate aminotransferase, and alkaline phosphatase.
In certain embodiments EM is defined as a patient whose overall survival (OS) is 12 weeks or less.
In one embodiment, any number of clinicopathological markers can be considered such as those typically measured to determine the relative disease state of a cancer patient and those typically measured to determine the relative health of an individual whether healthy or diseased.
In some embodiments, commercially available tests can be used to measure concentrations of clinicopathological markers, such as antibody-based assays where antibodies specific for a marker of interest are used to target such markers. Binding of the antibodies to the specific markers can be detected and concentrations and/or concentrations of the marker of interest can be determined by the relative amount of antibody binding. Additional tests can be employed, such as those that measure the relative abundance of specific cell type, for example, flow cytometers. Further tests that can be used herein include quantitative PCR, and similar technologies that provide information regarding relative gene expression concentrations. Additional tests are also contemplated herein as would be recognized by a skilled person for measuring various clinicopathological markers for a person's relative health or a specific disease state.
In some embodiments, samples used can include tissue biopsies, cerebrospinal fluid, lymph, whole blood, blood fractions, such as serum or specific blood cell populations, urine, sweat, tears, saliva, and/or feces.
In one embodiment, a patient can be assessed upon arrival at a clinic for enrollment in a clinical trial. In this case, a blood draw can be performed using standard techniques, and the patient's clinicopathological markers (e.g., neutrophils, hemoglobin, aspartate aminotransferase, alkaline phosphatase, etc.) can be assessed and the patient's overall survival predicted over a given period of time, such as 4, 8, 12 or 24 weeks, 4, 6, 8, or 10 months, 1, 2, 3, 4, or 5 years.
In another embodiment, a patient can be preapproved, remotely recruited, or flagged for accelerated therapy by having a lab report reviewed remotely where a centralized software platform can assess the patient's clinicopathological markers (e.g., neutrophils, hemoglobin, aspartate aminotransferase, alkaline phosphatase, etc.) and the overall survival determined.
In some embodiments, the present invention provides a method of conducting a clinical trial for an anticancer therapeutic. The method includes the steps of identifying candidate participants for a clinical trial, wherein the candidate participants have cancer, determining blood concentrations of i) neutrophils, ii) hemoglobin, iii) aspartate dehydrogenase, and iv) alkaline phosphatase of the candidate participants, calculating a predicted overall survival of the candidate participants, administering a first anticancer therapeutic to candidate participants with a predicted overall survival of 12 weeks or greater and administering a second anticancer therapeutic to candidate participants with a predicted overall survival of 12 weeks or less, and reducing at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of metastasis of the candidate participants.
In other embodiments, the invention provides a method of treating cancer in a subject in need thereof. The method includes the steps of taking a sample from the subject, determining concentrations of i) neutrophils, ii) hemoglobin, iii) aspartate dehydrogenase, and iv) alkaline phosphatase in the sample, calculating a predicted overall survival score of the subject, administering a first anticancer therapeutic to the subject if the subject's predicted overall survival is 12 weeks or greater or administering a second anticancer therapeutic to the subject if the subject's a predicted overall survival is less than 12 weeks, and reducing at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of metastasis in the subject.
In some embodiments, the present invention provides a method of conducting a clinical trial for an anticancer therapeutic. The method includes the steps of identifying candidate participants for a clinical trial, wherein the candidate participants have cancer, determining blood concentrations of i) neutrophil/lymphocyte ratio, ii) neutrophils, iii) albumin, iv) lactate dehydrogenase, v) gamma-glutamyltransferase, and vi) aspartate dehydrogenase of the candidate participants, calculating a predicted overall survival score of the candidate participants, administering a first anticancer therapeutic to candidate participants with a predicted overall survival of 12 weeks or greater and administering a second anticancer therapeutic to candidate participants with a predicted overall survival score 12 weeks or less, and reducing at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of metastasis of the candidate participants.
In other embodiments, the invention provides a method of treating cancer in a subject in need thereof. The method includes the steps of taking a sample from the subject, determining concentrations of i) neutrophil/lymphocyte ratio (NLR) proteins, ii) neutrophils, iii) albumin; iv) lactate dehydrogenase, v) gamma-glutamyltransferase, and vi) aspartate dehydrogenase in the sample, calculating a predicted overall survival of the subject, administering a first anticancer therapeutic to the subject if the subject's a predicted overall survival is 12 weeks or greater or administering a second anticancer therapeutic to the subject if the subject's predicted overall survival is less than 12 weeks, and reducing at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of metastasis in the subject.
In some embodiments where a patient is predicted to have an overall survival of less than 12 weeks from the date of assessment, a different, accelerated, and/or more aggressive treatment regimen can be prescribed than may have been intended for a particular clinical trial for which the patient was applying when assessed for overall survival.
Immune Checkpoint Inhibitors and Early MortalityAt the forefront of the ICI class are antibodies that target programmed cell death ligand-1 (PD-L1) exhibit therapeutic activity by restoring an antitumor T-cell response. An example of the profound impact demonstrated by ICIs is the 23% 5-year overall survival observed in PD-L1 high advanced or metastatic NSCLC patients treated by pembrolizumab (Garon, et al. 2019). However, during the development of these agents the EM phenomenon has been observed frequently when ICI were delivered as monotherapy in the metastatic disease setting, occurring in the early months of treatment in clinical trials across multiple tumor types including NSCLC (Borghaei, et al. 2015, Rizvi, et al. 2018, Socinski, et al. 2018, Mok, et al. 2019), UC (Bellmunt, et al. 2017, Powles, et al. 2018), HNSCC (Ferris, et al. 2016) and gastric cancer (Shitara, et al. 2018). Early mortality has been observed regardless of any specific antibody targeting PD-1 or PD-L1 (Winquist, et al. 2018, Mulkey, et al. 2019), suggesting a class effect. Though the phenomenon has been widely described, there remains a critical need for an objective measure or tool to better predict a patient's risk for EM and help direct treatment to patients most likely to have a favorable benefit:risk profile when being treated with ICI therapy.
Prevalence of Early Mortality Expectancy Across Clinical Trials Investigating ICITo ensure a comprehensive analysis across tumor types, the prevalence of the death within 4, 8 and 12 weeks of first study treatment was first evaluated in recent Phase I-III studies. These studies included:
-
- Study 1108 (CD-ON-MEDI4736-1108), A Phase 1 Study to Evaluate the Safety, Tolerability, and Pharmacokinetics of MEDI4736 in Subjects with Advanced Solid Tumors (n=1002)
- Study 06 (D419000006), A Phase 1b Open-label Study to Evaluate the Safety and Tolerability of MEDI4736 in Combination with Tremelimumab in Subjects with Advanced Non-small Cell Lung Cancer (n=379 pts)
- Study 10 (D4190000010), A Phase 1b/2 Multicenter, Open-label Study to Evaluate the Safety, Tolerability, and Efficacy of MEDI4736 in Combination with Tremelimumab in Subjects with Advanced Malignancies (n=168 pts)
- Study 21 (D4190000021), A Phase 1b/2 Study of MEDI4736 in Combination with Tremelimumab, MEDI4736 Monotherapy, and Tremelimumab Monotherapy in Subjects with Metastatic or Recurrent Gastric or Gastroesophageal Junction Adenocarcinoma (n=113 pts).
- CONDOR (D4193C00003), A Phase II, Multi-Center, Single-Arm, Global Study of MEDI4736 Monotherapy, Tremelimumab monotherapy, and MEDI4736 in combination with Tremelimumab in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=263).
- HAWK (D4193C00001), A Phase II, Multi-Center, Single-Arm, Global Study of MEDI4736 Monotherapy in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=112).
- ATLANTIC (D4191C00003), A Phase II, Non-comparative, Open Label, Multi-centre, International Study of MEDI4736, in Patients with Locally Advanced or Metastatic Non-Small Cell Lung Cancer (Stage IIIB-IV) Who Have Received at Least 2 Prior Systemic Treatment Regimens Including 1 Platinum-based Chemotherapy Regimen (N=371).
- MYSTIC (D419AC00001), A Phase III Randomized, Open-Label, Multi-Center, Global Study of MEDI4736 in Combination with Tremelimumab Therapy or MEDI4736 Monotherapy Versus Standard of Care Platinum-Based Chemotherapy in First Line Treatment of Patients with Advanced or Metastatic Non Small-Cell Lung Cancer (N=1118)
EAGLE (D4193C00002), A Phase III Randomized, Open-label, Multi-center, Global Study of MEDI4736 Monotherapy and MEDI4736 in Combination with Tremelimumab Versus Standard of Care Therapy in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=736)
By pooling these studies together, a range of 17.84% to 30.36% of patients treated by ICI were observed to have died ≤12 weeks after the first day of study treatment (Table 9).
Overall, the potential of higher EM in patients treated with ICI relative to active comparator signifies the need for improved prognostic methodologies to better identify those patients who are unlikely to survive beyond 12 weeks and thus benefit from ICI treatment.
Existing Prognostic Scores for Predicting Early MortalityA number of studies have been conducted to improve physicians' prognostic accuracy for patients with advanced or metastatic cancer (Ploquin, et al., Crit Rev Oncol Hematol. 2012) and have contributed to the development of prognostic scores for use in the clinic. Most of these studies identified specific clinical characteristics and laboratory values that are associated with poor outcomes in clinical studies. Several issues have limited the utility of these studies in addressing the early mortality phenomenon. First, several studies relied upon subjective and investigator-dependent parameters such as performance status (i.e., Eastern Cooperative Oncology Group (“ECOG”), Karnfosky index) (Krishnan M; J Support Oncol 2013). Another limitation of these clinical scores is the lack of validation in large independent studies. Finally, very few of these scores have been developed in the ICI era, so their relevance in this context remains unknown. A summary of these scores follows.
Royal Marsden Hospital (RMH) Score
The RMH score was amongst the first scores to predict early life expectancy in oncology. It was originally developed in the context of phase 1 trials investigating cytotoxic and targeted therapies. The RMH score is based on the following variables: lactate dehydrogenase (LDH, >upper limit of normal [ULN]), albumin (<35 g/L), and number of metastatic sites (>2) (Arkenau, et al., 2009). Each variable gives 1 point. A patient is classified poor prognostic by RMH if the sum of these variables exceeds 1. The application of the RMH score in large retrospective cohorts (N>1000 pts) of patients across tumor types (Wheler, et al. 2012) confirmed the poor prognosis associated with RMH score >1. High risk RMH was associated with poor prognosis in the ICI context in a retrospective series of patients (Bigot, et al., 2017; Minami et al, 2019).
Gustave Roussy Immune (GRIm) Score
The GRIm-Score is a more recent attempt to improve the life expectancy estimation of patients in the context of oncology phase 1 trials investigating ICI. It is based on albumin (<35 g/l), LDH (>ULN), and neutrophil to lymphocyte ratio (NLR; >6) (Bigot, et al. 2017). Similar to the RMH score, each variable gives 1 point. Patients treated with ICI whose sum of variables >1 are associated with poor overall survival in retrospective series of patients (Bigot, et al., 2017; Minami, et al., 2019).
Lung Immune Prognostic Index (LIPI) Score
The LIPI score was developed to predict the outcome in NSCLC patients treated with ICI in a variety of settings including routine clinical care, expanded access, compassionate-use programs, and clinical trials. It is based on the following variables: derived neutrophils/(leukocytes minus neutrophils) ratio (dNLR) >3 and LDH>ULN. Each variable gives 1 point. NSCLC patients whose sum of variables >1 are associated with poor prognosis per LIPI. In the original publication, the LIPI score was applied retrospectively in two independent cohorts of NSCLC patients treated with ICI (N=305 pts) or chemotherapy (N=162 pts) (Mezquita et al., 2018). The authors found that LIPI score >1 was associated with poor overall survival in ICI treated patients but not in chemotherapy treated patients. However, a subsequent analysis of LIPI performed by the FDA on pooled clinical trial data from studies evaluating 1368 second line metastatic NSCLC patients who received ICI and 1072 patients who received chemotherapy did not confirm these results (Kazandjian, et al. 2018). This study indicated that LIPI may exert a prognostic impact irrespective of therapeutic modalities (ICI or chemotherapy) for 2nd line metastatic NSCLC. Finally, another study investigated the LIPI score in the context of metastatic patients with various solid tumors enrolled in phase 1 trials. While this analysis was performed on a very heterogeneous cohort, a LIPI score >1 was also associated with poor overall survival, suggesting its prognostic role is not limited to NSCLC patients (Varga, et al. 2019).
Immune FitnessAll six 3i blood test variables (NLR, Neutrophils, Albumin, LDH, GGT, AST) assessed through standard laboratory measures and identified by feature selection, have previously been reported to be prognostic in various tumor types. For example, high NLR has been associated with poor prognosis in multiple tumor types including gastrointestinal (Bowen, et al. 2017), melanoma (Ding, et al. 2018), NSCLC (Fukui, et al. 2019), UC (Sacdalan, et al. 2018) and HNSCC (Yu, et al. 2018, Tham, et al. 2018). The ratio likely reflects a measure of immune fitness which may impact tumor immune responses, and the association with poor prognosis is judged to be independent of type of ICI. Fukui, et al. observed a significant association between high NLR and poor prognosis in NSCLC patients treated with nivolumab. In a meta-analysis of ICI studies in melanoma, NSCLC, and UC, Sacdalan, et al. observed an association of high NLR with poorer outcomes, suggesting it has potential as a prognostic marker. It has also been speculated that tumor burden and chronic inflammation lead to high NLR followed by poor prognosis (Bigot, et al. 2017). Absolute neutrophil count (ANC) is a surrogate marker of physiological stress including inflammation, and in NSCLC patients, high ANC after nivolumab treatment has been observed to be independently associated with inferior overall survival (Khunger, et al. 2018). Albumin (ALB) is a well-known marker of malnutrition and cachexia in cancer patients, and low ALB has been observed to be associated with poor overall survival in many malignancies such as SCCHN, breast, lung, and gastrointestinal cancers (Gupta, et al. 2010). The production of ALB is also reduced in the context of chronic inflammation.
LDH is an established marker that reflects increased cell turnover mainly in malignant disease, but also in non-malignant disease (e.g. hemolytic anemia). In a recent FDA analysis of early mortality in ICI studies, elevated LDH was reported as an important risk factor mainly in melanoma patients (Mulkey, et al. 2019). In a systematic review of 76 studies, high LDH was associated with poor prognosis in renal, melanoma, gastric, prostate, nasopharyngeal and lung cancers (Petrelli, et al. 2015). In advanced cancers, the increased concentrations of LDH in the blood reflect increased metabolic activity and rapid cell proliferation that result in increased leakage of the intracellular enzyme LDH into the blood circulation.
Abnormalities in liver function tests such as AST and GGT are carefully monitored in cancer patients. Abnormal liver function tests not only reflect pre-existing disease, but also cancer progression/liver metastasis and drug toxicity, which are all known causes of death among cancer patients. Hence, AST and GGT have been reported as prognostic markers in cancer (Freis, et al. 2017, Luo, et al. 2017), and GGT has also been suggested to play a role in tumor initiation, invasion and drug resistance.
Method of Identifying Immunocompromised PatientsThe 3i score can be used to identify patients with a high likelihood of EM, and who may therefore lack immune fitness. Lack of immune fitness suggests that ICI treatment should be postponed until such patients are effectively treated to improve their immune fitness. Therefore, therapeutic interventions in 3i high risk patients that bolster immune system fitness may enable these patients to receive to ICI treatment with a decreased risk of EM and moreover an increased chance for therapeutic benefit from an ICI treatment.
Similarly, methods of predicting the effectiveness of ICI treatment are contemplated herein. For example, a patient who is determined to be a 3i high risk patient would be predicted to not be effectively treated with an ICI, such as durvalumab.
Types of CancerTypes of cancer that are contemplated for treatment herein include, for example, NSCLC, advanced solid malignancies, biliary tract neoplasms, bladder cancer, colorectal cancer, diffuse large b-cell lymphoma, esophageal neoplasms, esophageal squamous cell carcinoma, extensive stage small cell lung cancer, gastric adenocarcinoma, gastric cancer, gastroesophageal junction cancer, head and neck cancer, head and neck squamous cell carcinoma, hepatocellular carcinoma, Hodgkin lymphoma, lung cancer, melanoma, mesothelioma, metastatic clear cell renal carcinoma, metastatic melanoma, metastatic non-cutaneous melanoma, multiple myeloma, nasopharyngeal neoplasms, non-Hodgkin lymphoma, ovarian cancer, fallopian tube cancer, peritoneal neoplasms, pleural mesothelioma, prostatic neoplasms, recurrent or metastatic PD-L1 positive or negative SCCHN, recurrent squamous cell lung cancer, renal cell cancer, renal cell carcinoma, SCCHN, hypo pharyngeal squamous cell carcinoma, laryngeal squamous cell carcinoma, small cell lung cancer, squamous cell carcinoma of the head and neck, squamous cell lung carcinoma, TNBC, transitional cell carcinoma, unresectable or metastatic melanoma, urothelial cancer, and urothelial carcinoma. In a particular embodiment, a treatment regimen contemplated can include a biological component, such as an antibody and a chemotherapeutic component. Contemplated antibodies include an anti-PD-L1 antibody such as durvalumab (MEDI4736), nivolumab, pembrolimumab, avelumab, atezolizumab, KNO35, an anti-PD-1 antibody such as REGN2810, SHR1210, IBI308, PDR001, Anti-PD-1, BGB-A317, BCD-100, and JS001, and an anti-CTLA4 antibody, such as tremelimumab or ipilimumab. Additional antibodies are also contemplated herein. Any therapeutically effective antibody subparts are also contemplated herein.
Methods of TreatmentIn one embodiment, the present disclosure is directed to effectively treating cancer patients with novel anticancer therapeutics and providing cancer patients with greater access to existing therapeutics. For example, the present disclosure contemplates treating cancer patients with immune checkpoint inhibitors (ICIs), such as those that target the PD-1/PD-L1 axis (PDX) and other IO treatments, such as immune system agonists.
Anticancer TherapeuticsInformation regarding durvalumab (or fragments thereof) for use in the methods provided herein can be found in U.S. Pat. Nos. 8,779,108; 9,493,565; and 10,400,039 the disclosures of which are incorporated herein by reference in its entirety. In a specific aspect, durvalumab or an antigen-binding fragment thereof for use in the methods provided herein comprises the variable heavy chain and variable light chain CDR sequences of the 2.14H9OPT antibody as disclosed in the aforementioned U.S. patents.
Information regarding tremelimumab (or antigen-binding fragments thereof) for use in the methods provided herein can be found in U.S. Pat. No. 6,682,736 (in which tremelimumab is referred to as 11.2.1), the disclosure of which is incorporated herein by reference in its entirety.
The amount of anticancer therapeutics, such as antibodies or antigen-binding fragments thereof to be administered to a patient will depend on various parameters such as the patient's age, weight, clinical assessment, tumor burden and/or other factors, including the judgment of the attending physician. Any acceptable route of administration is contemplated, such as, without limitation, administration intravenous (e.g., intravenous infusion), parenteral, or subcutaneous routes of administration.
Additional therapeutics (chemotherapies or biologics) contemplated herein include without limitation cisplatin/gemcitabine or methotrexate, vinblastine, ADRIAMYCIN™ (doxorubicin), cisplatin (MVAC), carboplatin-based regimen, or single-agent taxane or gemcitabine, temozolomide, or dacarbazine, vinflunine, docetaxel, paclitaxel, nab-paclitaxel, Vemurafenib, Erlotinib, Afatinib, Cetuximab, Bevacizumab, Erlotinib, Gefitinib, and/or Pemetrexed. Further examples include drugs targeting DNA damage repair systems, such as poly (ADP-ribose) polymerase 1 (PARP1) inhibitors and therapeutics inhibiting WEE1 protein kinase activity, ATR protein kinase activity, ATM protein kinase activity, Aurora B protein kinase activity, and DNA-PK activity.
Any therapeutic compositions or methods contemplated herein can be combined with one or more of any of the other therapeutic compositions and methods provided herein.
Companion DiagnosticsIn one embodiment, the present disclosure contemplates a device that includes one or more laboratory tests, as described elsewhere herein, for measuring concentrations of clinicopathological markers of a patient sample applied to the device. The device can further include a computing system comprising a gradient boosting machine learning model to predict whether the patient is likely to die within a given period of time based on the measured concentrations of clinicopathological markers from the one or more laboratory tests.
In another embodiment, a system is contemplated that includes one or more testing devices that is directly or indirectly linked to one or more computing devices such that upon determination of concentrations of one or more clinicopathological markers, the data be sent to a computing device to be provided to a gradient boosting machine learning model within the computing device. In turn, the gradient boosting machine learning model can output a prediction regarding patient EM and/or immune fitness or a prediction of effectiveness of an ICI therapeutic, such as durvalumab, to another component of the system.
In some embodiments, commercially available tests can be used to measure concentrations of clinicopathological markers, such antibody-based assays where antibodies specific for a marker of interest are used to target such markers. Binding of the antibodies to the specific markers can be detected and concentrations and/or concentrations of the marker of interest can be determined by the relative amount of antibody binding.
In some embodiments, samples used include tissue biopsies, cerebrospinal fluid, lymph, whole blood, blood fractions, such as serum or specific blood cell populations, urine, sweat, tears, saliva, and/or feces.
In one embodiment, a patient can be assessed upon arrival at a clinic for enrollment in a clinical trial. In this case, a blood draw can be performed using standard techniques and the patient's clinicopathological markers (e.g., neutrophils, hemoglobin, aspartate aminotransferase, alkaline phosphatase, etc.) can be assessed and the patient's overall survival predicted over a given period of time, such as 4, 8, 12 or 24 weeks, 4, 6, 8, or 10 months, 1, 2, 3, 4, or 5 years.
In another embodiment, a patient can be preapproved, remotely recruited, or flagged for accelerated therapy by having a lab report reviewed remotely where a centralized software platform can assess the patient's clinicopathological markers (e.g., neutrophils, hemoglobin, aspartate aminotransferase, alkaline phosphatase, etc.) and the overall score determined.
The following examples are set forth to provide those of ordinary skill in the art with a complete disclosure and description of how to practice the invention and are not intended to limit the scope of what the inventors regard as their invention.
Example 1—Validation of a Prognostic Score to Improve Patient Selection for Immuno-Oncology Trials (High Risk of Early Mortality; HREM)This example provides an overview of a new prognostic scoring methodology for use in immuno-oncology clinical trials. As used herein, “High Risk of Early Mortality”, “HREM” and “FastProgIO” are interchangeable.
BackgroundLife expectancy estimation is critical to selecting patients for oncology clinical trials, notably in the Phase 1 setting. However, most study protocols use subjective eligibility criteria to meet this goal (e.g., 12-week life expectancy). While eligibility criteria can help determine which patients could be included, they do not provide a reliable metric for determining overall survival (see
The purpose of this study was to develop a new prognostic score using basic laboratory tests at baseline to identify patients with high risk of early mortality (OS≤12 weeks) using data from three Phase I IO trials that included patients with non-small cell lung cancer (NSCLC) and urothelial carcinoma (UC). Further, this study also sought to compare the prevalence of patients with high risk of early mortality across clinical sites in these studies. The new prognostic score (HREM) was benchmarked against three published prognostic scores (RMH, GRIM and LIPI). Finally, the effects of tumour/immune cell programmed cell death ligand-1 (PD-L1) expression and liver metastasis on the prediction of patients with high risk of early mortality were assessed.
MethodsNSCLC and UC patients (N=972) prospectively enrolled in three Phase 1 trials (NCT01693562, NCT02000947, and NCT02261220) investigating durvalumab (anti-PD-L1)±tremelimumab (anti-cytotoxic T lymphocyte antigen [CTLA]-4) across 160 centers were analyzed. The variability in 12-week life expectancy rate across centers was also assessed.
A composite score (HREM) of clinicopathological variables predicting 12-week life expectancy was developed in 648 patients and validated in 324 patients using a multivariate regression method (
HREM performance (prediction of patients with high risk of early mortality with an overall survival of less than or equal to 12 weeks) was compared to that of existing published scores (see
Results
Fast progressor rates of patients analyzed per trial are indicated in Table 1 (see also,
Baseline characteristics for NSCLC and UC patients (shown in Table 2) were similar.
The HREM score included neutrophil counts, aspartate aminotransferase, alkaline phosphatase, and haemoglobin as the predictive markers. The true positive rate (TPR) of identifying patients with OS≤12 weeks was 74.7% (74.1% in the training data set and 75.9% in the testing data set), while the false positive rate (FPR) was controlled at around 10% (
Effect of PD-L1 Expression and Liver Metastasis in Predicted Patients with High Risk of Early Mortality
A PD-L1 effect was observed in patients with high risk of early mortality predicted by HREM but not in those predicted by the other scores: HREM predicted that patients with high risk of early mortality with PD-L1 Low/Neg expression progressed faster than those with PD-L1 High expression (
An effect of liver metastasis was observed in patients with high risk of early mortality predicted by all four scores: patients with liver metastasis (mets) progressed faster than those without (
Substantial variability across sites in the prevalence of patients with high risk of early mortality suggests that the use of subjective criteria to estimate 12-week life expectancy for study enrollment is clearly suboptimal. In this large cohort of NSCLC and UC patients, HREM significantly outperformed RMH, GRIM, and LIPI prognostic scores in the prediction of patients with high risk of early mortality. A PD-L1 effect was observed in patients with high risk of early mortality as predicted by HREM but not by the three published scores, which demonstrates the greater sensitivity of the HREM approach to assessing patients.
An effect of liver metastasis was observed in predicted patients with high risk of early mortality by all four scores. However, in patients without liver metastasis, patients with high risk of early mortality as predicted by HREM had a shorter OS than those predicted by the published scores. The success of the HREM prognostic assessment for NSCLC and UC patients suggests that this approach could be effective in additional cancer types.
Example 2—Evaluation of HREM in Various Tumour TypesThis example demonstrates the applicability of the HREM prognostic scoring methodology for use in scoring other cancer types.
MethodsPatients with various cancer types from a Phase I study (Study 1108) centers were analyzed for prevalence of predicted patients with high risk of early mortality.
ResultsFast progressor rates of patients analyzed are indicated in Table 3 (see also,
Table 4 illustrates a comparison of HREM versus other published scores.
HREM can be used to predict patients with high risk of early mortality in multiple cancer types. Therefore, this prognostic score can be used to improve patient intervention and phase I results in multiple cancer types. The strength of the prediction can be increased by increasing sample sizes.
Example 3—Correlation of HREM with Inflammatory ProteinsThis example investigated potential correlation between HREM results and inflammatory proteins.
MethodsUsing the UC cohort, patient serum samples were tested for concentrations of inflammatory proteins which were correlated with HREM scores using a Spearman correlation.
ResultsHREM scores were significantly correlated with 10 different inflammatory proteins (see
HREM could be useful for making therapeutically relevant biological inferences for that could help inform patient care and/or Phase I study selection.
Example 4—Development of a Pan-Tumor Prognostic Model (F1 Model)This example sought to develop an improved prognostic model (F1 model) building upon the success of the HREM model.
MethodsPatients from a combination of 4 studies with several cancer types were used to develop a new prognostic model with improved true positive and false positive rates (Table 5) (see also,
Development of the F1 model is illustrated in
Results
Predictive performance of the F1 model in the training and testing data is shown in
To demonstrate its usefulness in retrospective analysis of previous trials, the F1 model was used to review 2 phase I trials for gastroesophageal cancer using Durvalumab alone (Study 1108) or Durvalumab (D) alone, Tremelimumab (T) alone, Durvalumab+Tremelimumab (D+T), and Durvalumab+Tremelimumab in IFNy+cohorts (Study 21). Results are seen in comparing
ARCTIC study results (
The F1 model is an even more powerful predictor of OS that the HREM model. It outperforms published scores (poor OS, generalizable to all tumor types, PD-L1 effect). The F1 model can be used to appraise ongoing or previously conducted clinical trials in order to assess for any imbalance in the fast progressor rates between arms that may have potential impact on the outcome. Further, it is believed that the F1 model can be incorporated into inclusion/exclusion criteria in future phase I trials (escalation phase, randomized expansion) as a more objective way to exclude poor risk patients. Further, it can be used to improve the selection of study sites by excluding sites with history of enrolling high rate of patients with high risk of early mortality. Excluding patients with high risk of early mortality in phase 1 studies (escalation and randomized expansion parts) will likely increase the ability to detect efficacy signals of phase I trials.
Example 6—Correlation of F1 Model with Inflammatory ProteinsThis example investigated potential correlation between F1 model results and inflammatory proteins.
MethodsUsing the Study 1108, UC cohort, patient serum samples were tested for concentrations of inflammatory proteins (C-reactive protein, IL-6, TIMP-1, and IL-8) which were correlated with F1 model scores.
ResultsHREM scores were significantly correlated with 10 different inflammatory proteins (see
F1 model could be useful for making therapeutically relevant biological inferences for that could help inform patient care and/or Phase I study selection.
Example 7—Training and Testing of an Updated F1 Model Using the EAGLE Trial and Neutrophils (NEUT), Lymphocyte Percentage (LYM %), Albumin (ALB), Lactate Dehydrogenase (LDH), and URATE as Clinicopathological MarkersThis example sought to test an improved prognostic model (F1 model) building upon the success of the first F1 model.
Methods
Patients from a combination of 3 studies with several cancer types were used to develop a new prognostic model with improved true positive and false positive rates (Table 6).
Development of the model was performed as depicted in
To test the model, the EAGLE study was used. (736 participants with Squamous Cell Carcinoma of the Head and Neck, with three treatment schemes: chemotherapy, Durva Mono, and Durva+Treme). Overall survival for each scheme for intention-to-treat, PD-L1 high, and PD-L1 low in the trial is shown in
The updated F1 model predicted patients with high risk of early mortality (
This example sought to improve the F1 model by eliminating URATE from the clinicopathological markers used.
MethodsPatients from a combination of 3 studies with several cancer types were used to develop a new prognostic model with improved true positive and false positive rates (Table 7). The hazard ratio (HR) is calculated from proportional cox regression including PD-L1 (≥25% vs <25%), tumor location/HPV status and smoking status as strata. The p-value is calculated from a stratified log-rank test stratified by PD-L1 (≥25% vs <25%), tumor location/HPV status and smoking status.
Development of the model was performed as depicted in
To test the model, the EAGLE study was used. (736 participants with Squamous Cell Carcinoma of the Head and Neck, with three treatment schemes: chemotherapy, Durva Mono, and Durva+Treme).
ResultsThe F1 model using neutrophils (NEUT), lymphocyte percentage (LYM %), albumin (ALB), and lactate dehydrogenase (LDH) predicted patients with high risk of early mortality (
This example sought to use the findings from non-small cell lung cancer and squamous cell carcinoma head and neck cancer predictors of patients with high risk of early mortality to develop a single model for predicting both cancers. The non-small cell lung cancer markers evaluated are neutrophil/lymphocyte ratio (NLR); albumin (ALB), lactate dehydrogenase (LDH), gamma-glutamyltransferase (GGT), and aspartate aminotransferase (AST). The squamous cell carcinoma head and neck cancer predictors of patients with high risk of early mortality evaluated are neutrophils (NEUT), lymphocyte percentage (LYM %), albumin (ALB), lactate dehydrogenase (LDH), and URATE. Six markers were selected: neutrophil/lymphocyte ratio (NLR); neutrophils (NEUT), albumin (ALB), lactate dehydrogenase (LDH), gamma-glutamyltransferase (GGT), and aspartate aminotransferase (AST).
MethodsPatients from a combination of 7 studies with several cancer types were used to develop a new prognostic model with improved true positive and false positive rates (Table 8). The hazard ratio (HR) is calculated from proportional cox regression including PD-L1 (≥25% vs <25%), tumor location/HPV status and smoking status as strata. The p-value is calculated from a stratified log-rank test stratified by PD-L1 (≥25% vs <25%), tumor location/HPV status and smoking status.
Development of the model was performed as depicted in
To test the model, the ARCTIC, EAGLE (736 participants with Squamous Cell Carcinoma of the Head and Neck, with three treatment schemes: chemotherapy, Durva Mono, and Durva+Treme), and MYSTIC studies were used (1118 participants with PD-L1 high non-small cell lung cancer, with multiple treatment schemes: chemotherapy, Durva Mono, and Durva+Treme).
ResultsThe F1 model using neutrophil/lymphocyte ratio (NLR); neutrophils; albumin; lactate dehydrogenase; gamma-glutamyltransferase; and aspartate aminotransferase predicted patients with high risk of early mortality in the EAGLE trial (
The F1 model using neutrophil/lymphocyte ratio (NLR); neutrophils; albumin; lactate dehydrogenase; gamma-glutamyltransferase; and aspartate aminotransferase as clinicopathological markers is a powerful predictor of OS. The F1 model can be used to appraise ongoing or previously conducted clinical trials in order to assess for any imbalance in the fast progressor rates between arms that may have potential impact on the outcome. Further, it is believed that the F1 model can be incorporated into inclusion/exclusion criteria in future phase 1 trials (escalation phase, randomized expansion) as a more objective way to exclude poor risk patients. Further, it can be used to improve the selection of study sites by excluding sites with history of enrolling high rate of patients with high risk of early mortality. Excluding patients with high risk of early mortality in phase 1 studies (escalation and randomized expansion parts) will likely increase the ability to detect efficacy signals of phase I trials.
Example 10—Development of Immune Immediacy Index (3i Score) for Identifying Patients at Risk of Early Mortality SummaryNew Approach: Identify Patients with High Risk of Early Mortality in the ICI Context
While RMH, GRIM, and LIPI scores demonstrated prognostic impact, the recent availability of large set of patients treated with ICI in prospective trials at AstraZeneca represents an opportunity to improve the identification of patients with high risk of early mortality relative to an active chemotherapy comparator in the ICI context.
In randomized trials of ICI versus an active comparator in an advanced disease setting, where the crossing of the Kaplan-Meier (KM) curves for OS has been predominantly observed the higher rate of mortality in the ICI arm has been generally seen within the first 12 weeks after randomization. This is illustrated in the MYSTIC study of durvalumab versus chemotherapy in first line NSCLC.
Given the data collected across studies investigating ICI, the challenges in objectively predicting EM, and the limitations of previously established scores, we set out to use machine learning to develop a prognostic model (hereafter referred to as Immune Immediacy Index [3i score]) for identifying patients at risk of EM. Development workflow, including the definition of model requirements, design, development, and testing is described below.
Model Requirements
As illustrated in
The design choices in dataset identification for development, dataset separation for training, tuning and testing, as well as feature and method selection were driven by model requirements and cross-functional expertise in machine learning and clinical research.
Machine Learning Model Development Through Training, Tuning and TestingThe core activities in developing supervised machine learning models are most commonly sub-divided into three critical steps: training, tuning and testing.
During training, the model's parameters, such as tree depth, are optimized to encode relationships between input and output. In an iterative manner the model accepts an input from the training data set and produces an output which is compared against ground truth, and the error guides update of model parameters. Over time the model learns the representations of input that lead to desired target output using the training dataset.
The same machine learning technique can yield a different model in terms of architecture, weights and performance depending on what training constraints have been chosen. The optimal training constraints, called hyperparameters, are different in each machine learning situation and are set before training, as opposed to other model parameters that are developed during training.
Hyperparameter optimization, also called tuning, is a step in machine learning development that is often found to have a profound effect on a model's ability to generalize successfully. A choice of hyperparameters can lead to a model that will overfit or underfit on the same training set. One common approach to exploring a space of hyperparameter ranges is a grid search where each time a set of hyperparameters is sampled, the model is trained on a training data set and then evaluated on an independent dataset (tuning data set). Performance evaluation on a tuning set guides the selection of optimal training constraints that produce a model robust enough to function beyond the training set.
Model testing is performed to evaluate the final model fit on the training dataset using hyperparameters of choice. Testing is performed on an independent dataset, which was neither used in training nor in tuning, to evaluate how well the model generalises and derives patterns beyond the data it has encountered before.
Dataset IdentificationA comprehensive set of ICI trials was utilized with individual patient concentration data (Table 9) for developing the 3i score model. The trials were selected so that the training and tuning sets would include data for patients with a diverse range of tumor types and across different lines of therapy.
Dataset Separation for Training, Tuning and TestingData separation for training, tuning, testing was as follows:
Training set (used in final input feature selection, model training and hyperparameter optimization):
Study 1108 (N=923), Study 06 (N=353), Study 10 (N=164), Study 21 (N=98)
ATLANTIC (D4191C00003), A Phase II, Non-comparative, Open Label, Multi-centre, International Study of MEDI4736, in Patients with Locally Advanced or Metastatic Non-Small Cell Lung Cancer (Stage IIIB-IV) Who Have Received at Least 2 Prior Systemic Treatment Regimens Including 1 Platinum-based Chemotherapy Regimen (N=371)
CONDOR (D4193C00003), A Phase II, Multi-Center, Single-Arm, Global Study of MEDI4736 Monotherapy, Tremelimumab monotherapy, and MEDI4736 in combination with Tremelimumab in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=193)
HAWK (D4193C00001), A Phase II, Multi-Center, Single-Arm, Global Study of MEDI4736 Monotherapy in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=111)
Note: Patients censored before 12 weeks were not included in the training set.
Tuning set (used in model hyperparameter optimization):
MYSTIC (D419AC00001), A Phase III Randomized, Open-Label, Multi-Center, Global Study of MEDI4736 in Combination with Tremelimumab Therapy or MEDI4736 Monotherapy Versus Standard of Care Platinum-Based Chemotherapy in First Line Treatment of Patients with Advanced or Metastatic Non Small-Cell Lung Cancer (N=1118)
Test set (used in model testing):
EAGLE (D4193C00002), A Phase III Randomized, Open-label, Multi-center, Global Study of MEDI4736 Monotherapy and MEDI4736 in Combination with Tremelimumab Versus Standard of Care Therapy in Patients with Recurrent or Metastatic Squamous Cell Carcinoma of the Head and Neck (N=736)
The total training data set includes 2213 patients, including mostly patients diagnosed with NSCLC, SCCHN, UBC, and Gastric cancer, summarized in Table 10.
One strength of using certain machine learning algorithms is their ability to determine feature importance and weigh predictive features over non-predictive ones. Given that model complexity increases with a growing number of features or dimensions, and that one of the requirements was to use routinely available laboratory measurements (i.e. laboratory measurements covering hematology, chemistry and thyroid hormones available in the clinical trials selected here), a comprehensive starting list of 37 candidate predictors were collated and summarized in Table 11. Removal of non-important or highly-correlated features was planned to occur during the development stage.
Machine Learning Method Selection (Gradient Boosting Trees)
A gradient boosting modelling approach was chosen for predicting 12-week EM. The particular version of gradient boosting method adopted is called XGBoost (Extreme Gradient Boosting). This machine learning approach has increasingly become a preference of the statistical community because of its robust prediction performance, its speed (i.e. scalability on parallel processes), and accommodation of missing values (Friedman, et al. 2000, Friedman, et al. 2001, Chen and Guestrin 2016). Decision trees are used as a weak learner in gradient boosting. One variable is selected at each time to divide a node into two sub-nodes until no split is available to further improve the model fitting. A separate tree is built on the residuals of the previous tree. A variable may be selected multiple times within one tree and in different trees. Decision trees are sequentially built until stop criteria are met. The final gradient boosting model is built by assembling decision trees additively, possibly applying a weighted learning rate to each. Through selecting proper model hyperparameters, XGBoost models can be trained to minimize the prediction error with the controlled model complexity, which leads to a good balance of bias-variance trade-off. XGBoost model treats missing values by learning default directions. Internally, XGBoost learns the direction to go when a value is missing.
Test PlanThe test plan was designed to evaluate the 3i score by retrospective application on an ICI clinical trial test set (EAGLE). EAGLE study was excluded from training and tuning to represent an independent test set where the model would be evaluated in terms of (1) ability to generalize beyond training set as defined by true positive rate (TPR), false positive rate (FPR) and partial area under the curve (AUC) in specificity region between 0.7 and 0.9; (2) comparison to other, previously developed risk scores; (3) clinical utility as defined by increased therapeutic effect after exclusion of the 3i score high-risk group.
Code tests were not applicable in the testing plan as no new functionality was introduced, nor any existing functionality extended in open-source R packages used in development.
DevelopmentDevelopment Environment
All development was performed in R v.3.5.1 on x86_64 pc linux gnu(64-bit) operating system. All open source packages used for development are listed in Table 12.
Acquired datasets were pre-processed internally according to internal data management standards. Patients who died within 12 weeks (i.e. time from randomization/first dose to death ≤12 weeks) were labelled as EM. Patients who survived beyond 12 weeks (i.e. time from randomization/first dose to death >12 weeks) were not labelled as EM. Patients who were censored within 12 weeks (i.e. patients lost to follow up ≤12 weeks) were not included in the training set during the development of the model. No other data manipulations were performed.
Feature SelectionFeature selection was performed to examine a broad range of candidate predictors for the 3i score model, and to return a subset of predictors from which a performant model could be fitted. First, a full model was fitted using a gradient boosting fitting procedure on the training data (
A gradient boosting model was fitted to the training data to calculate variable importance. Variable importance is quantified by how much improvement in accuracy was brought by including the variable. All available pre-treatment blood test variables (n=31) covering haematology, clinical chemistry, and thyroid hormones were initially included in this first step (i.e., all laboratory variables). Clinical variables well-recognized to be associated with prognosis in oncology such as age (“AGE”), gender (“SEX”), baseline tumor size (estimated by RECIST, “TUMSZINV”), Body Mass Index (“BMI”), tumor type (“typ”), and the presence of liver metastasis (“LIVERBL”) were also introduced in this model. As shown in
All six blood test variables assessed through standard laboratory measures and identified by feature selection, have previously been reported to be prognostic in various tumor types.
High NLR has been associated with poor prognosis in multiple tumor types including gastrointestinal (Bowen, et al. 2017), melanoma (Ding, et al. 2018), NSCLC (Fukui, et al. 2019), UC (Sacdalan, et al. 2018) and HNSCC (Yu, et al. 2018, Tham, et al. 2018). The ratio likely reflects a measure of immune fitness which may impact tumor immune responses, and the association with poor prognosis is judged to be independent of type of ICI. Fukui, et al. observed a significant association between high NLR and poor prognosis in NSCLC patients treated with nivolumab. In a meta-analysis of ICI studies in melanoma, NSCLC and UC, Sacdalan, et al. observed an association of high NLR with poorer outcomes, suggesting it has potential as a prognostic marker. It has also been speculated that tumor burden and chronic inflammation lead to high NLR followed by poor prognosis (Bigot, et al. 2017).
Absolute neutrophil count (ANC) is a surrogate marker of physiological stress including inflammation, and in NSCLC patients, high ANC after nivolumab treatment has been observed to be independently associated with inferior overall survival (Khunger, et al. 2018).
ALB is a well-known marker of malnutrition and cachexia in cancer patients, and low ALB has been observed to be associated with poor overall survival in many malignancies such as SCCHN, breast, lung, and gastrointestinal cancers (Gupta, et al. 2010).
LDH is an established marker that reflects increased cell turnover mainly in malignant disease, but also in non-malignant disease (eg, haemolytic anaemia). In a recent FDA analysis of early mortality in ICI studies, elevated LDH was reported as an important risk factor mainly in melanoma patients (Mulkey, et al. 2019). In a systematic review of 76 studies, high LDH was associated with poor prognosis in renal, melanoma, gastric, prostate, nasopharyngeal and lung cancers (Petrelli, et al. 2015). In advanced cancers, the increased concentrations of LDH in the blood reflect increased metabolic activity and rapid cell proliferation that result in increased leakage of the intracellular enzyme LDH into the blood circulation.
Abnormalities in liver function tests such as AST and GGT are carefully monitored in cancer patients. Abnormal liver function tests not only reflect pre-existing disease, but also cancer progression/liver metastasis and drug toxicity, which are all known causes of death among cancer patients. Hence, AST and GGT have been reported as prognostic markers in cancer (Freis, et al. 2017, Luo, et al. 2017), and GGT has also been suggested to play a role in tumor initiation, invasion and drug resistance.
Training and Hyperparameter Selection for the 3i Score ModelOnly the most predictive biological variables (NLR, neutrophils, albumin, LDH, AST, GGT), described in the feature selection step previously (
The final 3i score model, selected from the pool of 432 models, had pAUC(0.7, 0.9) of 0.83 (ranked No. 8 among all 432 models) in the training dataset and was trained with the following non-default XGBoost hyperparameters:
Maximum depth of a tree (max.depth): 3
Minimum loss reduction required to make a further partition on a leaf node of the tree (gamma): 1
Minimum sum of instance weight (hessian) needed in a child (min.child.weight):
Subsample ratio of columns when constructing each tree (colsample.bytree): 1
Subsample ratio of the training instances (subsample): 0.6
scale_pos_weight: negative_cases/postive_cases
nrounds: 22
To describe the nature of the final model, the importance of each variable selected in the final model was computed. As shown in
The following six key predictors: NLR, NEUT, ALB, LDH, GGT, AST, and the tumor type were retained in the final model. The model produces a score (i.e. value between 0 and 1) representing the probability of death occurring in ≤12 weeks for each patient. The 3i score is then converted to a status that assigns patients to prognostic or risk categories of high or low. The cut-off was determined to allow 10% false positive rate at predicting EM in the training data set and calculated as 0.649 on 2 Jul. 2019. Patients with a score above the cut-off of 0.649 are identified as high risk of EM and patients at or below the cut-off (0.649) are identified as low risk of EM. For patients missing any of the 6 lab test values, the 3i score will not be calculated. This decision was supported by the expectation that globally, patients will have data on these variables readily available as they are routinely-collected, standard laboratory measures.
3i Vs GRIm and LIPI in Training and Tuning Data SetsWhen applied in the training set (N=2213 patients), the TPR at 12 weeks was 67% and the FPR was 10%. Performance as defined by pAUC (0.7, 0.9) was 0.83. The median OS of patients identified with a high risk 3i score status (486/2213 patients) was 9.29 weeks (95% CI: 8.29 weeks, 9.86 weeks) and the median OS of patients with a low risk 3i score status was 61.43 weeks (95% CI: 57.71, 66.14) (see
The patients identified as high risk by the 3i score (see
When applied in the tuning set (MYSTIC) (N=1118) (see
In the MYSTIC intent-to-treat population (ITT) (see
In the MYSTIC PD-L1≥25% subgroup (see
When applied in the test set (EAGLE) (N=736) (see
In the EAGLE ITT (see
In the PD-L1≥25% subgroup (N=140) (see
In the MYSTIC and EAGLE studies, a Wilcoxon test was used to compare the 6 lab values used in the 3i score between patients identified with a high risk (predicted life expectancy ≤12 weeks) or low risk (predicted life expectancy of >12 weeks) by the 3i score. For all 6 parameters, the values from patients with a high risk and low risk 3i score status are significantly different (p<0.05) (see
The variables included in the 3i score are associated with host immune fitness and cancer-related inflammation. Further, the additional translational analyses conducted from samples acquired in clinical trials show that 3i high-risk patients are characterized with higher baseline concentration of blood peripheral immune suppressive cells (MDSC) and higher expression of immune suppressive gene signatures and pro-inflammatory protein concentrations in the serum (CRP, IL6 and IL8). Interestingly, unlike immune checkpoint blockade, chemotherapy is well recognized to reduce the IL6, IL8, and CRP concentrations. This finding could explain the early mortality phenomenon observed with ICB vs. chemotherapy in patients with advanced/metastatic cancer.
To investigate whether the 3i score can be used as an indicator of patient immune fitness, the baseline molecular profiles of 3i high-risk patients were assessed by blood gene expression, peripheral blood immunoprofiling by flow cytometry, and serum protein expression.
Results Peripheral Blood Gene Expression3i high (poor prognosis) is associated with lower baseline T effector and higher immune-suppressive cell gene expression signatures compared to 3i low in NSCLC, HNSCC, and UBC patients irrespective of patient PD-L1 status (see
3i high is associated with lower baseline T, NK, and B cells and higher monocytic myeloid-derived suppressive cells (MNC MDSC) absolute counts compared to 3i low in NSCLC patients (see
3i high (poor prognosis) is highly associated with higher baseline pro-inflammatory protein concentrations in particular C-reactive protein (CRP) and interleukins IL6 and IL8 compared with 3i low NSCLC, HNSCC, and UBC patients (see
The 3i index appears to have clinical utility in identifying patients who may lack immune fitness. Lack of immune fitness suggests that ICI treatment should be postponed until such patients are effectively treated to improve their immune fitness or that patients should be treated with a non-ICI treatment. Therefore, therapeutic interventions in 3i high patients that bolster immune system fitness may enable these patients to receive to ICI treatment with a decreased risk of EM and moreover an increased chance for therapeutic benefit from ICI treatment.
Example Computing SystemsIn this example, computing device 7000 includes processor 7002, memory 7004, network interface 7006, and input/output unit 7008, all of which may be coupled by system bus 7010 or a similar mechanism. In some embodiments, computing device 7000 may include other components and/or peripheral devices (e.g., detachable storage, printers, and so on).
Processor 7002 may be one or more of any type of computer processing element, such as a central processing unit (CPU), a co-processor (e.g., a mathematics, graphics, or encryption co-processor), a digital signal processor (DSP), a network processor, and/or a form of integrated circuit or controller that performs processor operations. In some cases, processor 7002 may be one or more single-core processors. In other cases, processor 7002 may be one or more multi-core processors with multiple independent processing units. Processor 7002 may also include register memory for temporarily storing instructions being executed and related data, as well as cache memory for temporarily storing recently-used instructions and data.
Memory 7004 may be any form of computer-usable memory, including but not limited to random access memory (RAM), read-only memory (ROM), and non-volatile memory (e.g., flash memory, hard disk drives, solid state drives, compact discs (CDs), digital video discs (DVDs), and/or tape storage). Thus, memory 7004 represents both main memory units, as well as long-term storage. Other types of memory may include biological memory.
Memory 7004 may store program instructions and/or data on which program instructions may operate. By way of example, memory 7004 may store these program instructions on a non-transitory, computer-readable medium, such that the instructions are executable by processor 7002 to carry out any of the methods, processes, or operations disclosed in this specification or the accompanying drawings.
As shown in
Network interface 7006 may take the form of one or more wireline interfaces, such as Ethernet (e.g., Fast Ethernet, Gigabit Ethernet, and so on). Network interface 7006 may also support communication over one or more non-Ethernet media, such as coaxial cables or power lines, or over wide-area media, such as Synchronous Optical Networking (SONET) or digital subscriber line (DSL) technologies. Network interface 7006 may additionally take the form of one or more wireless interfaces, such as IEEE 802.11 (Wifi), BLUETOOTH®, global positioning system (GPS), or a wide-area wireless interface. However, other forms of physical layer interfaces and other types of standard or proprietary communication protocols may be used over network interface 7006. Furthermore, network interface 7006 may comprise multiple physical interfaces. For instance, some embodiments of computing device 7000 may include Ethernet, BLUETOOTH®, and Wifi interfaces.
Input/output unit 7008 may facilitate user and peripheral device interaction with computing device 7000. Input/output unit 7008 may include one or more types of input devices, such as a keyboard, a mouse, a touch screen, and so on. Similarly, input/output unit 7008 may include one or more types of output devices, such as a screen, monitor, printer, and/or one or more light emitting diodes (LEDs). Additionally or alternatively, computing device 7000 may communicate with other devices using a universal serial bus (USB) or high-definition multimedia interface (HDMI) port interface, for example.
In some embodiments, one or more computing devices like computing device 7000 may be deployed to support a machine leaning architecture. The exact physical location, connectivity, and configuration of these computing devices may be unknown and/or unimportant to client devices. Accordingly, the computing devices may be referred to as “cloud-based” devices that may be housed at various remote data center locations.
For example, server devices 7102 can be configured to perform various computing tasks of computing device 7000. Thus, computing tasks can be distributed among one or more of server devices 7102. To the extent that these computing tasks can be performed in parallel, such a distribution of tasks may reduce the total time to complete these tasks and return a result. For purposes of simplicity, both server cluster 7100 and individual server devices 7102 may be referred to as a “server device.” This nomenclature should be understood to imply that one or more distinct server devices, data storage devices, and cluster routers may be involved in server device operations.
Data storage 7104 may be data storage arrays that include drive array controllers configured to manage read and write access to groups of hard disk drives and/or solid state drives. The drive array controllers, alone or in conjunction with server devices 7102, may also be configured to manage backup or redundant copies of the data stored in data storage 7104 to protect against drive failures or other types of failures that prevent one or more of server devices 7102 from accessing units of data storage 7104. Other types of memory aside from drives may be used.
Routers 7106 may include networking equipment configured to provide internal and external communications for server cluster 7100. For example, routers 7106 may include one or more packet-switching and/or routing devices (including switches and/or gateways) configured to provide (i) network communications between server devices 7102 and data storage 7104 via local cluster network 7108, and/or (ii) network communications between server cluster 7100 and other devices via communication link 7110 to network 7112.
Additionally, the configuration of routers 7106 can be based at least in part on the data communication requirements of server devices 7102 and data storage 7104, the latency and throughput of the local cluster network 7108, the latency, throughput, and cost of communication link 7110, and/or other factors that may contribute to the cost, speed, fault-tolerance, resiliency, efficiency, and/or other design goals of the system architecture.
As a possible example, data storage 7104 may include any form of database, such as a structured query language (SQL) database. Various types of data structures may store the information in such a database, including but not limited to tables, arrays, lists, trees, and tuples. Furthermore, any databases in data storage 7104 may be monolithic or distributed across multiple physical devices.
Server devices 7102 may be configured to transmit data to and receive data from data storage 7104. This transmission and retrieval may take the form of SQL queries or other types of database queries, and the output of such queries, respectively. Additional text, images, video, and/or audio may be included as well. Furthermore, server devices 7102 may organize the received data into web page or web application representations. Such a representation may take the form of a markup language, such as the hypertext markup language (HTML), the extensible markup language (XML), or some other standardized or proprietary format. Moreover, server devices 7102 may have the capability of executing various types of computerized scripting languages, such as but not limited to Perl, Python, PHP Hypertext Preprocessor (PHP), Active Server Pages (ASP), JAVASCRIPT®, and so on. Computer program code written in these languages may facilitate the providing of web pages to client devices, as well as client device interaction with the web pages. Alternatively or additionally, JAVA® may be used to facilitate generation of web pages and/or to provide web application functionality.
Example OperationsBlock 7200 of
Block 7202 may involve training, based on the model training data, a sequential series of decision trees until a stopping condition is reached, wherein each subsequent decision tree of the sequential series of decision trees is constructed based on residual values of its preceding decision tree of the sequential series of decision trees.
Block 7204 may involve generating a gradient boosting machine learning model as an additive function of the sequential series of decision trees, wherein the gradient boosting machine learning model predicts whether a further cancer patient dies within the threshold number of weeks based on the additive function applied to results from the laboratory tests as conducted on the further cancer patient and a tumor type of the further cancer patient.
In some embodiments, the threshold number of weeks is 12 weeks. But other values are possible (e.g., 8, 10, 15, or 20 weeks).
In some embodiments, the results from the laboratory tests provide one or more of: a neutrophil to lymphocyte ratio, a neutrophil concentration, an albumin concentration, a lactate dehydrogenase concentration, an aspartate amino transferase concentration, or a gamma-glutamyl transferase concentration. Some or all of these results may be used.
In some embodiments, the neutrophil to lymphocyte ratio, the neutrophil concentration, the albumin concentration, the lactate dehydrogenase concentration, the aspartate amino transferase concentration, and the gamma-glutamyl transferase concentration were selected from 20 or more laboratory tests conducted on the first set of cancer patients. The selecting may involve (i) training the gradient boosting machine learning model on the 20 or more laboratory tests; (ii) for each respective laboratory test, training a respective variation of the gradient boosting machine learning model on the 20 or more laboratory tests with the respective laboratory test removed; and (iii) determining predictive values of each of the 20 or more laboratory tests by comparing accuracies of the variations of the gradient boosting machine learning model. The 20 or more laboratory tests may be selected from the predictors of Table 11, and may include non-test predictors, such as gender and age.
In some embodiments, the tumor type is one of: non-small cell lung cancer, urothelial carcinoma, squamous cell carcinoma of head and neck, gastric cancer, renal cell carcinoma, or Hodgkin's lymphoma. Other types may be possible.
In some embodiments, the additive function applies a multiplicative learning rate to outcomes from each of the sequential series of decision trees.
In some embodiments, predicting whether the further cancer patient dies within the threshold number of weeks comprises determining, by using the additive function, a probability that the further cancer patient dies within the threshold number of weeks.
In some embodiments, predicting whether the further cancer patient dies within the threshold number of weeks further comprises based on the probability and a cutoff value, placing the further cancer patient into either a high risk of death category or a low risk of death category. In some embodiments, the cutoff value is selected so that there is no more than a false positive rate of 5-20% when predicting whether the first set of cancer patients die within the threshold number of weeks. In some cases, the cutoff value may be selected so that the false positive rate is 10%.
In some embodiments, cancer patients placed in the high risk of death category are excluded from a clinical trial for treatment of cancer, and wherein cancer patients placed in the low risk of death category are included in the clinical trial for treatment of cancer.
In some embodiments, the stopping condition is that the sequential series of decision trees includes a threshold number of decision trees. Alternatively or additionally, the stopping condition is that the residual values of a particular decision tree in the sequential series of decision trees are all below a threshold value.
Some embodiments may further involve obtaining model tuning data, wherein the model tuning data involves a second set of cancer patients that underwent cancer treatments, wherein the model tuning data associates (i) results from the laboratory tests as conducted on the second set of cancer patients and tumor types of the second set of cancer patients with (ii) whether the second set of cancer patients died within the threshold number of weeks from initiation of the cancer treatments; generating two or more variations of the gradient boosting machine learning model by training the sequential series of decision trees for each variation with different sets of hyperparameters; applying the variations to the model training data and the model tuning data; and selecting, as the gradient boosting machine learning model, a particular variation from the variations that provides predictions of whether the first set of cancer patients and the second set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments within a threshold degree of accuracy.
The threshold degree of accuracy can be based on a true positive rate or a true negative rate of predictions of whether the first set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments. Alternatively or additionally, the threshold degree of accuracy can be based on a p-value for interaction between: (i) predictions of whether the second set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments, and (ii) PD-L1 positive patients in the second set of cancer patients.
In some embodiments, the hyperparameters include one or more of: maximum depth of the decision trees set to 2, 3, 4, 5, or 6, minimum loss reduction required to split a node in the decision trees set to 0, 1, 2, 3, or 4, minimum sum of weights of all observations required in a child node in the decision trees set between 3.0 and 7.0, fraction of input parameters to be randomly sampled for each of the decision trees set between 0.7 and 1.0, fraction of observations to be randomly sampled for each of the decision trees set between 0.4 and 0.8, number of decision trees in the sequential series of decision trees set between 10 and 25. Other ranges may be used as well.
Block 7300 of
Block 7302 may involve obtaining results from the laboratory tests as conducted on a further cancer patient.
Block 7304 may involve obtaining a tumor type of the further cancer patient.
Block 7306 may involve applying the gradient boosting machine learning model to results from the laboratory tests as conducted on a further cancer patient and the tumor type of the further cancer patient.
Block 7308 may involve receiving, from the gradient boosting machine learning model, a prediction of whether the further cancer patient dies within the threshold number of weeks.
In some embodiments, the threshold number of weeks is 12 weeks. But other values are possible (e.g., 8, 10, 15, or 20 weeks).
In some embodiments, the results from the laboratory tests provide one or more of: a neutrophil to lymphocyte ratio, a neutrophil concentration, an albumin concentration, a lactate dehydrogenase concentration, an aspartate amino transferase concentration, or a gamma-glutamyl transferase concentration. Some or all of these results may be used.
In some embodiments, the tumor type is one of: non-small cell lung cancer, urothelial carcinoma, squamous cell carcinoma of head and neck, gastric cancer, renal cell carcinoma, or Hodgkin's lymphoma. Other types may be possible.
In some embodiments, the additive function applies a multiplicative learning rate to outcomes from each of the sequential series of decision trees.
In some embodiments, predicting whether the further cancer patient dies within the threshold number of weeks comprises determining, by using the additive function, a probability that the further cancer patient dies within the threshold number of weeks.
In some embodiments, predicting whether the further cancer patient dies within the threshold number of weeks further comprises based on the probability and a cutoff value, placing the further cancer patient into either a high risk of death category or a low risk of death category. In some embodiments, the cutoff value was selected so that there is no more than a false positive rate of 5-20% when predicting whether the first set of cancer patients die within the threshold number of weeks. In some cases, a cutoff value may be selected so that the false positive rate is 10%.
In some embodiments, cancer patients placed in the high risk of death category are excluded from a clinical trial for treatment of cancer, and wherein cancer patients placed in the low risk of death category are included in the clinical trial for treatment of cancer.
In some embodiments, the stopping condition is that the sequential series of decision trees includes a threshold number of decision trees. Alternatively or additionally, the stopping condition is that the residual values of a particular decision tree in the sequential series of decision trees are all below a threshold value.
In some embodiments, the gradient boosting machine learning model was also trained using model tuning data, wherein the model tuning data involves a second set of cancer patients that underwent cancer treatments, wherein the model tuning data associates (i) results from the laboratory tests as conducted on the second set of cancer patients and tumor types of the second set of cancer patients with (ii) whether the second set of cancer patients died within the threshold number of weeks from initiation of the cancer treatments, and wherein training the gradient boosting machine learning model using the model tuning data comprises: (i) generating two or more variations of the gradient boosting machine learning model by training the sequential series of decision trees for each variation with different sets of hyperparameters; (ii) applying the variations to the model training data and the model tuning data; and (iii) selecting, as the gradient boosting machine learning model, a particular variation from the variations that provides predictions of whether the first set of cancer patients and the second set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments within a threshold degree of accuracy.
The threshold degree of accuracy can be based on a true positive rate or a true negative rate of predictions of whether the first set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments. Alternatively or additionally, the threshold degree of accuracy can be based on a p-value for interaction between: (i) predictions of whether the second set of cancer patients die within the threshold number of weeks from initiation of the cancer treatments, and (ii) PD-L1 positive patients in the second set of cancer patients.
In some embodiments, the hyperparameters include one or more of: maximum depth of the decision trees set to 2, 3, 4, 5, or 6, minimum loss reduction required to split a node in the decision trees set to 0, 1, 2, 3, or 4, minimum sum of weights of all observations required in a child node in the decision trees set between 3.0 and 7.0, fraction of input parameters to be randomly sampled for each of the decision trees set between 0.7 and 1.0, fraction of observations to be randomly sampled for each of the decision trees set between 0.4 and 0.8, number of decision trees in the sequential series of decision trees set between 10 and 25. Other ranges may be used as well.
ConclusionA novel model has been developed that relies on baseline measures, as opposed to subjective clinical measures like ECOG scores. Based on the evaluation of this index in the MYSTIC and EAGLE studies, this index appears to have clinical utility in identifying patients who may have a relatively more favorable benefit:risk profile compared to the ITT population.
All patents and publications mentioned in this specification are herein incorporated by reference to the same extent as if each independent patent and publication was specifically and individually indicated to be incorporated by reference. Citation or identification of any reference in any section of this application shall not be construed as an admission that such reference is available as prior art to the present invention.
REFERENCES
- Arkenau H T, Barriuso J, Olmos D, Ang J E, de Bono J, Judson I, et al. Prospective validation of a prognostic score to improve patient selection for oncology phase I trials. J Clin Oncol. 2009; 27(16):2692-6.
- Bellmunt J, de Wit R, Vaughn D J, Fradet Y, Lee J L, Fong L et al. Pembrolizumab as Second-Line Therapy for Advanced Urothelial Carcinoma. N Engl J Med. 2017; 376(11):1015-26.
- Bigot F, Castanon E, Baldini C, Hollebecque A, Carmona A, Postel-Vinay S, et al. Prospective validation of a prognostic score for patients in immunotherapy phase I trials: The Gustave Roussy Immune Score (GRIm-Score). Eur J Cancer. 2017; 84:212-8.
- Borghaei H, Paz-Ares L, Horn L, Spigel D R, Steins M, Ready N E, et al. Nivolumab versus Docetaxel in Advanced Nonsquamous Non-Small-Cell Lung Cancer. N Engl J Med. 2015; 373(17):1627-39.
- Bowen R C, Little NAB, Harmer J R, Ma J, Mirabelli L G, Roller K D, et al. Neutrophil-to-lymphocyte ratio as prognostic indicator in gastrointestinal cancers: a systematic review and meta-analysis. Oncotarget. 2017; 8(19):32171-89.
- Caliskan, A, Bryson J J, and Narayanan A. Semantics derived automatically from language corpora contain human-like biases. Science. 356(6334):183-186, 2017.
- Champiat S, Ferrara R, Massard C, Besse B, Marabelle A, Soria J C, et al. Hyperprogressive disease: recognizing a novel pattern to improve patient management. Nat Rev Clin Oncol. 2018; 15(12):748-62.
- Chen T, Guestrin C. XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD). 2016; 785-794. DOI: http://dx.doi.org/10.1145/2939672.2939785.
- Chen T, He T, Benesty M, Khotilovich V, Tang Y, Cho H, Chen K, Mitchell R, Cano I, Shou T, Li M, Xie J, Lin M, Geng Y, Li Y. R Package ‘xgboost’, Mar. 12, 2019. https://cran.r-project.org/web/packages/xgboost/xgboost.pdf.
- Ding Y, Zhang S, Qiao J. Prognostic value of neutrophil-to-lymphocyte ratio in melanoma: Evidence from a PRISMA-compliant meta-analysis. Medicine (Baltimore). 2018; 97(30):e11446.
- Ferris R L, Blumenschein G, Jr., Fayette J, Guigay J, Colevas A D, Licitra L, et al. Nivolumab for Recurrent Squamous-Cell Carcinoma of the Head and Neck. N Engl J Med. 2016; 375(19):1856-67.
- Friedman J, Hastie T, Tibshirani R. (2000). Additive logistic regression: a statistical view of boosting (with discussion and a rejoinder by the authors). The annals of statistics. 2000; 28(2), 337-407.
- Friedman J H. Greedy function approximation: a gradient boosting machine. Annals of Statistics, pp. 2001; 1189-1232.
- Freis P, Graillot E, Rousset P, Hervieu V, Chardon L, Lombard-Bohas C, et al. Prognostic factors in neuroendocrine carcinoma: biological markers are more useful than histomorphological markers. Sci Rep. 2017; 7:40609.
- Fukui T, Okuma Y, Nakahara Y, Otani S, Igawa S, Katagiri M, et al. Activity of Nivolumab and Utility of Neutrophil-to-Lymphocyte Ratio as a Predictive Biomarker for Advanced Non-Small-Cell Lung Cancer: A Prospective Observational Study. Clin Lung Cancer. 2019; 20(3):208-14 e2.
- Garon E B, Hellmann M D, Rizvi N A, Carcereny E, Leighl N B, Ahn M J, et al. Five-Year Overall Survival for Patients With Advanced NonSmall-Cell Lung Cancer Treated With Pembrolizumab: Results From the Phase I KEYNOTE-001 Study. J Clin Oncol. 2019; JC01900934.
- Gupta D, Lis C G. Pretreatment serum albumin as a predictor of cancer survival: a systematic review of the epidemiological literature. Nutr J. 2010; 9:69.
- Kazandjian D G, Gong Y, Kazandjian H, Pazdur R, Blumenthal G M. Exploration of baseline derived neutrophil to lymphocyte ratio (dNLR) and lactate dehydrogenase (LDH) in patients (pts) with metastatic non-small cell lung cancer (mNSCLC) treated with immune checkpoint inhibitors (ICI) or cytotoxic chemotherapy (CCT). 2018; 36(15_suppl):3035.
- Khunger M, Patil P D, Khunger A, Li M, Hu B, Rakshit S, et al. Post-treatment changes in hematological parameters predict response to nivolumab monotherapy in non-small cell lung cancer patients. PLoS One. 2018; 13(10):e0197743.
- Luo M, Sun W, Wu C, Zhang L, Liu D, Li W, et al. High pretreatment serum gamma-glutamyl transpeptidase predicts an inferior outcome in nasopharyngeal carcinoma. Oncotarget. 2017; 8(40):67651-62.
- Mezquita L, Auclin E, Ferrara R, Charrier M, Remon J, et al. Association of the Lung Immune Prognostic Index With Immune Checkpoint Inhibitor Outcomes in Patients With Advanced Non-Small Cell Lung Cancer. JAMA Oncol. 2018 Mar. 1; 4(3):351-357.
- Minami S, Ihara S, Ikuta S, Komuta K. Gustave Roussy Immune Score and Royal Marsden Hospital Prognostic Score Are Biomarkers of Immune-Checkpoint Inhibitor for Non-Small Cell Lung Cancer. World J Oncol. 2019 April; 10(2):90-100.
- Mok T S K, Wu Y L, Kudaba I, Kowalski D M, Cho B C, Turna H Z, et al. Pembrolizumab versus chemotherapy for previously untreated, P D-L1-expressing, locally advanced or metastatic non-small-cell lung cancer (KEYNOTE-042): a randomised, open-label, controlled, phase 3 trial. Lancet. 2019; 393(10183):1819-30.
- Mulkey F, By K, Theoret M R, Maher V E, Pazdur R, Sridhara R. Analysis of early mortality in randomized clinical trials evaluating anti-PD-1/PD-L1 antibodies: A systematic analysis by the United States Food and Drug Administration (FDA). 2019; 37(15_suppl):2516-.
- Petrelli F, Cabiddu M, Coinu A, Borgonovo K, Ghilardi M, Lonati V, et al. Prognostic role of lactate dehydrogenase in solid tumors: a systematic review and meta-analysis of 76 studies. Acta Oncol. 2015; 54(7):961-70.
- Powles T, Duran I, van der Heijden M S, Loriot Y, Vogelzang N J, De Giorgi U et al. Atezolizumab versus chemotherapy in patients with platinum-treated locally advanced or metastatic urothelial carcinoma (IMvigor211): a multicentre, open-label, phase 3 randomised controlled trial. Lancet. 2018; 391(10122):748-57.
- Rizvi N, Cho B C, Reinmuth N, Lee K H, Ahn M J, Luft A et al. Durvalumab with or without tremelimumab vs platinum-based chemotherapy as first-line treatment for metastatic non-small cell lung cancer: MYSTIC. ESMO Immuno-Oncology Congress; 2018 Dec. 13-16; Geneva, Switzerland. Abstract LBA6 [oral presentation].
- Sacdalan D B, Lucero J A, Sacdalan D L. Prognostic utility of baseline neutrophil-to-lymphocyte ratio in patients receiving immune checkpoint inhibitors: a review and meta-analysis. Onco Targets Ther. 2018; 11:955-65.
- Socinski M A, Jotte R M, Cappuzzo F, Orlandi F, Stroyakovskiy D, Nogami N, et al. Atezolizumab for First-Line Treatment of Metastatic Nonsquamous NSCLC. N Engl J Med. 2018; 378(24):2288-301.
- Shitara K, Ozguroglu M, Bang Y J, Di Bartolomeo M, Mandala M, Ryu M H, et al. Pembrolizumab versus paclitaxel for previously treated, advanced gastric or gastro-oesophageal junction cancer (KEYNOTE-061): a randomised, open-label, controlled, phase 3 trial. Lancet. 2018; 392(10142):123-33.
- Tham T, Bardash Y, Herman S W, Costantino P D. Neutrophil-to-lymphocyte ratio as a prognostic indicator in head and neck cancer: A systematic review and meta-analysis. Head Neck. 2018; 40(11):2546-57.
- Varga A, Bernard-Tessier A, Auclin E, Mezquita Perez L. Baldini C, et al. Applicability of the Lung Immune Prognostic Index (LIPI) in patients with metastatic solid tumors when treated with Immune Checkpoint Inhibitors (ICI) in early clinical trials. Ann Oncol. 2019 Feb. 1; 30(Supplement 1).
- Wheler J, Tsimberidou A M, Hong D, Naing A, Falchook G, Piha-Paul S, et al. Survival of 1,181 patients in a phase I clinic: the M D Anderson Clinical Center for targeted therapy experience. Clin Cancer Res. 2012; 18(10):2922-9.
- Winquist E K S, Nichols A C, Lenehan J G, Macneil S D, Palma D A; Early mortality with immune checkpoint inhibitors (IOs) in solid tumors: An inconvenient truth? J Clin Oncol. 2018; 36 (suppl; abstr 12121).
- Yu Y, Wang H, Yan A, Wang H, Li X, Liu J, et al. Pretreatment neutrophil to lymphocyte ratio in determining the prognosis of head and neck cancer: a meta-analysis. BMC Cancer. 2018; 18(1):383.
Claims
1. A method of treating a patient having cancer, comprising:
- a) determining concentrations of multiple clinicopathological markers of the patient;
- b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers;
- c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has a likelihood of dying within a period of 12 weeks; and
- d) administering an anticancer therapeutic to the patient if the prediction indicates that the patient does not have a likelihood of dying within the period of 12 weeks.
2. The method of claim 1, wherein the multiple clinicopathological markers comprise one or more of age, albumin (g/L), alanine phosphatases (U/L), alanine amino transferase (U/L), aspartate amino transferase (U/L), basophils (103/μL), basophils/Leukocytes (%), bilirubin (μmol/L), Body Mass Index, calcium (mmol/L), chloride (mmol/L), eosinophils (103/μL), eosinophils/leukocytes (%), gamma-glutamyl transferase (U/L), glucose (mmol/L), hematocrit (%), hemoglobin (g/L), potassium (mmol/L), lactate dehydrogenase (U/L), liver metastasis (yes/no), lymphocytes (103/μL), lymphocytes/leukocytes (%), magnesium (mmol/L), monocytes (103/μL), monocytes/leukocytes (%), neutrophils (103/μL), neutrophils/leukocytes (%), neutrophils/lymphocytes (%), platelets (103/μL), protein (g/L), gender, sodium (mmol/L), thyrotropin (mU/L), tumor size, uric acid (mmol/L), and leukocytes (103/μL).
3. The method of claim 2, wherein the multiple clinicopathological markers comprise neutrophils/leukocytes (%), neutrophils (103/μL), albumin (g/L), lactate dehydrogenase (U/L), aspartate amino transferase (U/L), and/or gamma-glutamyl transferase (U/L).
4. The method of claim 1 further comprising taking a sample from the patient.
5. The method of claim 4, wherein the sample comprises a tissue biopsy, cerebrospinal fluid, lymph, whole blood, serum, a blood cell, urine, sweat, tears, saliva, and/or feces.
6. The method of claim 5, wherein a concentration of one or more of the multiple clinicopathological markers is determined from the sample taken from the patient.
7. The method of claim 1, wherein the anticancer therapeutic comprises cisplatin, gemcitabine, methotrexate, vinblastine, doxorubicin, cisplatin (MVAC), carboplatin, a taxane, temozolomide, dacarbazine, vinflunine, docetaxel, paclitaxel, nab-paclitaxel, Vemurafenib, Erlotinib, Afatinib, Cetuximab, Bevacizumab, Gefitinib, and/or Pemetrexed.
8. The method of claim 1, wherein the anticancer therapeutic comprises an immune checkpoint inhibitor.
9. The method of claim 8, wherein the immune checkpoint inhibitor comprises an anti-CTLA-4 antibody, an anti-PD-1 antibody, and/or an anti-PD-L1 antibody.
10. The method of claim 9, wherein the anti-CTLA-4 antibody is tremelimumab or ipilimumab.
11. The method of claim 9, wherein the anti-PD-1 antibody is REGN2810, SHR1210, IBI308, PDR001, nivolumab, pembrolizumab Anti-PD-1, BGB-A317, BCD-100, or JS001.
12. The method of claim 9, wherein the anti-PD-L1 antibody comprises durvalumab, avelumab, atezolizumab, or KNO35.
13. The method of claim 1 further comprising determining a cancer type of the patient and also providing, to the gradient boosting machine learning model, the determined cancer type.
14. A method of treating a patient having cancer, comprising:
- a) determining concentrations of multiple clinicopathological markers of the patient;
- b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers;
- c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has poor immune fitness; and
- d) administering an anticancer therapeutic to the patient if the prediction indicates that the patient does not have poor immune fitness.
15. The method of claim 14 further comprising e) reducing in the patient at least one of a rate of cancer cell division, tumor growth, tumor size, tumor density, or rate of tumor metastasis.
16. A method of treating a patient having cancer, comprising:
- a) determining concentrations of multiple clinicopathological markers of the patient;
- b) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers;
- c) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has poor immune fitness;
- d) administering a therapeutic to the patient to boost the patient's immune fitness if the prediction indicates that the patient has poor immune fitness;
- e) determining second concentrations of multiple clinicopathological markers of the patient;
- f) providing, to a gradient boosting machine learning model, the determined second concentrations of the multiple clinicopathological markers;
- g) receiving, from the gradient boosting machine learning model, a second prediction of whether the patient has poor immune fitness; and
- h) administering an anticancer therapeutic to the patient if the prediction indicates that the patient does not have poor immune fitness.
17. A method of conducting a clinical trial for an anticancer therapeutic, comprising:
- a) identifying a candidate participant for a clinical trial, wherein the candidate participant has cancer;
- b) determining concentrations of multiple clinicopathological markers of the patient;
- c) providing, to a gradient boosting machine learning model, the determined concentrations of the multiple clinicopathological markers;
- d) receiving, from the gradient boosting machine learning model, a prediction of whether the patient has a likelihood of dying within a period of 12 weeks; and
- e) including the patient in the clinical trial for the anticancer therapeutic if the prediction indicates that the patient does not have a likelihood of dying within the period of 12 weeks.
18. The method of claim 17, wherein the anticancer therapeutic is an immune checkpoint inhibitor.
Type: Application
Filed: Oct 18, 2019
Publication Date: Jun 18, 2020
Inventors: LI YU (GAITHERSBURG, MD), HARRY YANG (GAITHERSBURG, MD), MOHAMMED DAR (GAITHERSBURG, MD), LORIN ROSKOS (GAITHERSBURG, MD), JEAN-CHARLES SORIA (GAITHERSBURG, MD), CHARLES FERTE (GAITHERSBURG, MD), WEI ZHAO (GAITHERSBURG, MD), ALINE GENDRIN BROKMANN (CAMBRIDGE), JOLYON FARIA (CAMBRIDGE), PRALAY MUKHOPADHYAY (WILMINGTON, DE)
Application Number: 16/657,359