System and method for linking patient level molecular data to treatment selection

Abstract

Pulmonary arterial hypertension (PAH) exhibits an obliterative vasculopathy where complex, integrated pathobiological signaling pathways drive vascular remodeling. In PAH, the arteriopathy includes numerous endophenotypes that occur to differing extents across patients. Variability in the proteomic and genetic profile is observed, causing phenotypic heterogeneity and inconsistent clinical responses to drug therapies. We have used network medicine to discover modifiable therapeutic targets in PAH by generating patient-specific protein-protein interaction (PPI) networks to unmask molecular interactions that identify and distinguish groups of individual patients with the same clinical phenotype. This allows personalized clinical phenotyping in PAH in those patient groups. The findings here also clarify the relationship between PAH genetic risk and pathobiology on an individual patient level, and inform treatment rationales and personalized drug selection using the PPI networks. Overall, findings from this project will advance precision medicine in PAH with direct relevance to the clinical management of patients.

Description

Attorney Docket No. 15024-398PC0 SYSTEM AND METHOD FOR LINKING PATIENT LEVEL MOLECULAR DATA TO TREATMENT SELECTION CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application is an international PCT application which claims the benefit of United States provisional application serial no. 63 / 648,947, filed 17 May 2024. The entire contents of the aforementioned application is hereby incorporated by reference as if fully set forth herein. GOVERNMENT FUNDING SUPPORT

[0002] This invention was made with government support under grant nos. HL155096, HL141268, HL174007, and GM103652, awarded by the National Institutes of Health. The government has certain rights in the invention. BACKGROUND

[0003] 1. Field of the Invention

[0004] This invention relates to the general field of medicine and more specifically to the field of precision or personalized medicine useful in diseases and conditions where patients exhibit variable treatment response patterns. The method takes molecular heterogeneity and reduces this information to patient-specific data that can be used to select a treatment or treatments that are most likely to be successful.

[0005] 2. Background of the invention

[0006] Pulmonary arterial hypertension (PAH) is a rare, progressive, and fatal disease with numerous (>14) orphan drugs approved for its treatment. Multidrug regimens are now recommended as the standard of care to minimize the risk of a poor outcome, but long-term morbidity and mortality remains poor in PAH (60% mortality at 5 years). More than 16 medications spanning multiple drug classes are now approved for PAH, and there is increasing emphasis on goal-oriented, combination therapy. Approved therapies, while effective, are costly (> $0.5 million / year per patient when used in combination), burdensome, and not associated with improved health-related quality of life.

[0007] PAH is a highly morbid cardiopulmonary disease characterized by an obliterative vasculopathy involving severe distal pulmonary arteriopathy that promotes right heart failure. Despite a uniform clinical definition, inconsistent treatment response to the same drug or drug class is reported commonlyAttorney Docket No. 15024-398PC0 in clinical trials and at point-of-care. This has been ascribed to widely variable pathobiology across individual PAH patients, which causes differences in disease characteristics and clinical profile. Thus, analytical strategies that individualize the molecular signature of PAH patients are needed. Personalizing PAH pathobiology, in turn, can establish a key framework to permit optimization of PAH phenotyping and prognostication, as well as tailor therapy selection. Using patient-specific pathogenetic information to inform drug development, clinical trial design, and treatment selection in clinical practice may improve efficacy and effectiveness of PAH drugs, respectively, but such strategies have been lacking. Progress in achieving this goal has been hindered by challenges inherent to sampling affected tissue in PAH and the availability of analytical methods that align individual patient-level pathobiological data to treatment.

[0008] PAH is now recognized as a heterogeneous disease with multiple endophenotypes. There has been a call for precision-based approaches to treat this complex and fatal disease to improve outcomes and minimize burden for both patients and the healthcare system. Complex and integrated pathobiological signaling pathways drive vascular remodeling in PAH. The arteriopathy includes numerous endophenotypes (i.e., specific features, such as fibrosis, cellular proliferation, hypertrophy, others) that occur to differing extent across patients. Interplay between numerous endophenotypes underlies the arterial remodeling pattern in PAH. A hallmark PAH endophenotype is pulmonary artery endothelial cell (PAEC) dysfunction; ex vivo studies of PAECs from PAH patients validate numerous disease-specific pharmacotherapeutic targets experimentally, but these samples derive largely from end- stage patients at the time of lung explant.

[0009] Wide variability in the proteomic and genetic profile is observed in PAH, which accounts for phenotypic heterogeneity and inconsistent clinical response to drug therapies reported in clinical trials and at point-of-care. These observations suggest that an opportunity exists to improve clinical outcome by individualizing the pathobiology-clinical phenotype relationship. Thus, utilizing methods that illustrate overlapping and integrated signaling pathways is a particularly attractive approach to understanding PAH.

[0010] Precision medicine allows personalization of treatments for individuals in a population to ensure that each treatment group will receive an effective treatment. This method is designed to optimize the efficiency and therapeutic benefits for certain patients, using genetic or molecular profiling. In this way, medical treatments can be specifically tailored to a patient’s individual characteristics such as geneticAttorney Docket No. 15024-398PC0 background, environment, and lifestyle, with the aim of delivering the right treatment to the right patient at the right time.

[0011] Nonetheless, precision medicine in PAH remains unrealized. SUMMARY OF THE INVENTION

[0012] Thus, there is a need in the art for a new approach to develop precision medicine, for PAH and other diseases. Here, the approach focuses on protein-protein interactions (PPI) to predict clinical characteristics, outcome, and optimal drug selection in individual PAH patients. A minimally invasive “cell biopsy” approach in which PAECs can be collected from the tips of pulmonary artery catheters and propagated in vitro is discussed here. Thus, right heart catheterization (RHC) with associated PAEC biopsy provides a direct window into the lung vasculature by sampling the principle tissue compartment involved in PAH pathogenesis during a procedure that is required for diagnosis, drug development, and clinical trials, and fundamental to the longitudinal management of patients.

[0013] We recently innovated a network medicine method using transcriptomic data to generate individualized interactomes (also referred to as individualized PPIs) that emphasize functionally relevant protein-protein interactions (PPIs). This approach captured patient-specific molecular features (that are otherwise marginalized when using reductionist analytical methods), which could then be used to predict clinical traits in cardiomyopathy patients. The method uses a transcriptome-interactome-systems pharmacology bioinformatics pipeline in which PAECs collected at point-of-care function as the pathobiological substrate for analysis. We used this approach and implemented a blinded clinical trial design in silico. The proportion of therapeutic targets of the prescribed PAH therapy found in each cell- biopsy individualized interactome is associated with improved clinical outcomes in PAH. Findings from this study have advanced precision medicine by linking patient level molecular data to treatment selection and, ultimately, therapeutic efficacy and clinical effectiveness that can be useful in various diseases.

[0014] In particular, the invention relates to a method of monitoring a complex disease defined by molecular heterogeneity across patients sharing the same diagnosis in a patient in need thereof, comprising: (a) obtaining a cell biopsy sample from tissue affected by the disease in the subject, wherein the sample is cells or a cell surrogate; (b) expanding cells in the cell biopsy sample in tissue culture;Attorney Docket No. 15024-398PC0 (c) isolating patient mRNA from the expanded cells and subjecting the mRNA to bulk RNA- SEQ; (d) compiling the transcript quantity for each gene in the patient mRNA to form a patient transcript profile and holding the profile for later analysis; (e) isolating control mRNA from cells of a normal control and subjecting the mRNA to bulk RNA-SEQ; (f) assembling the transcript quantity for each gene in each control subject in a matrix and calculating the Pearson correlation coefficient for each gene pair in the matrix; (g) adding the patient transcript profile to the matrix of (f); (h) recalculating the Pearson correlation coefficient for each gene pair in the matrix; (i) gathering the gene pairs with Pearson correlation coefficients that are statistically significantly different after the addition of the patient transcript profile to the matrix; (j) mapping the gene pairs of (i) to a consolidated human protein-protein interactome (PPI); (k) assembling the gene pairs that map to protein-protein interactions in the consolidated human PPI to produce a patient-specific PPI; and (l) monitoring the disease-specific gene expression and the clinical changes of the disease in the subject.

[0015] In certain embodiments, the cell surrogate is selected from the group consisting of peripheral blood, serum, urine, cerebrospinal fluid, and saliva. In certain embodiments, the complex disease is a pulmonary vascular disease. In certain embodiments, the pulmonary vascular disease is pulmonary arterial hypertension.

[0016] In other embodiments, the cell biopsy sample is a sample of pulmonary artery endothelial cells, and can be obtained by right heart catheterization (RHC) and harvesting from the tips of RHC balloons.

[0017] In other embodiments, the complex disease is selected from the group consisting of cardiomyopathy; parenchymal lung disease; solid tumors affecting the breast, lung, colon, pancreas, or ovaries; inflammatory bowel disease; systemic lupus erythematosus and systemic sclerosis; mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, and metabolic dysfunction; and the like. In some embodiments, the complex disease is Crohn’s disease or ulcerative colitis.

[0018] In certain embodiments, the disease-specific gene expression signature and clinical changes in the subject are monitored to provide real-time information of the disease pathobiology.Attorney Docket No. 15024-398PC0

[0019] The invention further provides a method of treating a complex disease defined by molecular heterogeneity across patients sharing the same diagnosis in a subject in need thereof, comprising: (a) performing (a) through (k) of the above method; (b) compiling a list of the molecular targets of the disease; (c) aligning the molecular targets of the disease with the individualized PPI of the patient to find overlap between the disease targets and the patient PPI; (d) treating the patient with a treatment modality that overlaps with the patient’s individualized PPI.

[0020] In specific embodiments of the method, the complex disease is a pulmonary vascular disease. Preferably, the pulmonary vascular disease is pulmonary arterial hypertension.

[0021] In some embodiments, the individualized PPI are patient- and time-specific wiring maps.

[0022] In other embodiments, the complex disease is selected from the group consisting of cardiomyopathy; parenchymal lung disease; solid tumors affecting the breast, lung, colon, pancreas, or ovaries; inflammatory bowel disease; systemic lupus erythematosus and systemic sclerosis; mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, and metabolic dysfunction; and the like. In certain embodiments, the complex disease is Crohn’s disease or ulcerative colitis. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Certain embodiments are illustrated by way of example, and not by way of limitation in the figures of the accompanying drawings.

[0024] FIG.1A, FIG.1B, and FIG.1C are a set of schematic drawings.

[0025] FIG.2 is a set of diagrams that show the progression of pulmonary arterial hypertension.

[0026] FIG.3 is a schematic for the study described herein. RHC=right heart catheterization; 6MWD=six-minute walk distance; BNP=brain natriuretic peptide; REVEAL=Registry to Evaluate Early and Long-Term PAH Disease Management; PAEC=pulmonary artery endothelial cell; PAH=pulmonary arterial hypertension; PPI=protein-protein interaction.

[0027] FIG.4A and FIG.4B provide a schematic of the methods of the invention.

[0028] FIG.5 is a schematic showing steps of the analytical pipeline.

[0029] FIG.6 presents flow cytometry data confirming endothelial cell lineage.

[0030] FIG.7A- 7E is a set of photomicrographs showing stained biopsy material as indicated.Attorney Docket No. 15024-398PC0

[0031] FIG. 8 provides information on the steps taken for individualizing the pathobiology of a disease condition such as PAH.

[0032] FIG. 9A and FIG. 9B are charts showing the study design for groups A and B.

[0033] FIG. 10 is a table providing information on patients’ number of proteins and number of interactions.

[0034] FIG. 11A and FIG. 11B are heatmaps showing results for node overlap and edge overlap similarity among the individual PAH networks, respectively.

[0035] FIG. 12 is a multidimensional scaling plot of the PAEC RNA-Seq dataset.

[0036] FIG. 13 presents the distribution of the coefficient of variance (CV) in the expression of genes among PAH samples. PAEC=pulmonary arterial endothelial cells; PAH=pulmonary arterial hypertension.

[0037] FIG. 14 shows sample correlations among PAH participant and non-diseased control samples.

[0038] FIG. 15 is an illustration of some patient-specific networks.

[0039] FIG. 16A and FIG. 16B are heatmaps showing the node (FIG. 16A) and edge (FIG. 16B) overlap similarity among individual PAH networks.

[0040] FIG. 17 is a heatmap of endophenotype enrichment in PAH patient samples.

[0041] FIG. 18A is a heatmap of the proportion of targets within a drug class expressed by each participant. FIG. 18B is a graph showing the proportion of participants expressing each target.

[0042] FIG. 19A through FIG. 19F show the association between PAH therapeutic target-interactome alignment and six-minute walk distance (6MWD) (FIG. 19A and FIG. 19B); brain natriuretic peptide (GNP) levels (FIG. 19C and FIG. 19D) and functional class (FIG. 19E and FIG. 19F) at six and twelve months after right heart catheterization.

[0043] FIG. 20 is a set of graphs that show the association between PAH therapeutic target-interactome alignment and improvement in the United States Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) 2.0 risk score (see FIG. 20A and FIG. 20B) and the Comparative, Prospective Registry of Newly Initiated Therapies for PH (COMPERA) 2.0 four-strata risk assessment (see FIG. 20C and FIG. 20D) at six and twelve months after right heart catheterization (RHC), respectively.

[0044] FIG. 21A and FIG. 21B are plots showing the relationship between serotonin-psychotropic comparator-target alignment and the U.S. Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) 2.0 risk scores at six and 12 months, respectively.Attorney Docket No. 15024-398PC0

[0045] FIG.22A and FIG.22B are diagrams showing the temporal change of the patient-specific networks over different samples

[0046] FIG.23A through FIG.23D are heatmap diagrams. Row=drug class. Columns=serial pulmonary artery endothelial cell sampling via RHC. Cell = proportion of targets expressed at the time of RHC. Black = 100% targets expressed. Peach = 0% targets expressed. sGC = stimulator of soluble guanylate cyclase. DHEA = dehydroepiandrosterone. RHC = right heart catheterization. PAH = pulmonary arterial hypertension. IPAH = idiopathic PAH. PDE5i = phosphodiesterase 5 inhibitor. ERA = endothelin receptor antagonist.

[0047] FIG.24A and FIG.24B are graphs showing the REVEAL 2.0 Risk Score results at six (FIG. 24A) and 12 (FIG.24B) months after RHC.

[0048] FIG.25A, FIG.25B, and FIG.25C are graphs showing drug-interactome pairs that predict PAH treatment response.

[0049] FIG.26A through FIG.26F are graphs showing that drug-interactome pairs predict PAH treatment response.

[0050] FIG.27A through FIG.27B are schematic diagrams. FIG 27A shows significant correlations in patients. FIG.27B is the proteomic profile of an individual patient. FIG.27C shows steps for determining significant correlations.

[0051] FIG.28A-28B shows the variance in the transcriptomic profile of patients with hypertrophic cardiomyopathy compared to controls.

[0052] FIG.29A presents a set of data for individualized interactomes for each HCM patient. Nodes represent proteins, and edges represent functional protein-protein interactions (PPIs). FIG.29B shows that there was 73% overlap in network nodes across HCM patients, but only 48% overlap for PPIs.

[0053] FIG.30 is a heatmap identifying 25 different new HCM endophenotypes.

[0054] FIG.31 is a bar graph showing that the proportion of HCM patients with networks enriched for fibrosis signaling pathways was significantly greater.

[0055] FIG.32A shows the number of fibrosis nodes in edges unique to a single HCM patient network, provided in descending order. FIG.32B is a graph showing a positive correlation between the number of unique fibrosis edges for each patient network and interstitial collagen quantity.Attorney Docket No. 15024-398PC0 DETAILED DESCRIPTION OF THE INVENTION

[0056] 1. Overview

[0057] Translational approaches and therapeutic pipelines in PAH treatment have been limited by (1) a lack of direct access to tissues or cells from the diseased organ of interest (the lungs), (2) the study of tissues only from end-stage disease at death or explant or from outside of the diseased organ of interest, (3) a lack of monitoring methods for endophenotypic and molecular changes over the course of disease in living patients and (4) there is no way to actively monitor the pulmonary vascular disease signature in response to therapeutic changes.

[0058] To this end, we have reported that physical protein-protein interaction (PPI) networks enriched for gene ontology pathways provide key data on the pathogenetic basis of PAH endophenotypes. Studying functionally significant signaling pathways using network medicine is designed to discover novel and modifiable therapeutic targets in PAH. This approach differs from classic reductionist methods that infer functionality based on transcript or protein quantity alone, which may erroneously implicate molecular bystanders in the pathogenesis of disease. This method involves a precision-based network medicine strategy that generates patient-specific PPI networks (e.g., patient-level molecular wiring map). Our approach unmasks molecular interactions that distinguish (and group together) individual patients with the same clinical phenotype.

[0059] The data presented here show that developing patient-specific PPI networks can personalize clinical phenotyping and optimize prognosis in PAH. These findings also clarify the relationship between PAH genetic risk and pathobiology on an individual-patient level, and inform rationale and personalized drug selection using the PPI networks.

[0060] The dataset used here includes comprehensive proteomic, genomic, clinical, and outcome data across two timepoints (1 year apart) for idiopathic PAH, hereditary PAH, asymptomatic family members of patients with PAH, including three patients that developed PAH during the study, and healthy volunteer controls (N=500 total). The testing was designed to produce (1) profile patient-specific PPI networks using proteomic and genetic data, and analyze temporal differences in network features by patient group and (2) develop, test, and validate a network score that informs phenotype and outcome of individual PAH patients. The PPI networks are used to predict patient-specific drug therapies. Findings from this project will advance precision medicine in PAH with direct relevance to the clinical management of patients.Attorney Docket No. 15024-398PC0

[0061] Manipulation of large amounts of data is needed to create precision medicine treatments. In general, the genetics of an individual and the biological pathways of the disease are studied to find the functional relevance of the genetics to the biological pathways. Next, the functional pathways and nuanced phenotyping are linked to find individualized coupling of pathobiology and phenotype. Finally, the drug selection for the patient is tailored to the coupled pathobiology-phenotype to produce an optimized clinical treatment. See FIG. 1.

[0062] In a variety of heart diseases, including pulmonary arterial hypertension (PAH), fibrosis and remodeling of lung vasculature is a shared development of the various pathophysiologies. Systemic sclerosis or systemic scleroderma, is an autoimmune or connective tissue disease, characterized by thickening of the skin caused by accumulation of collagen, and by injuries to the smallest arteries. PAH is heterogenous and has several barriers to individualized treatment. First, biological data from affected tissue in PAH for clinical use are lacking. Second, analytic methodologies that leverage, rather than isolate, individual patient-level molecular data have not been used in PAH. See FIG. 2 for a chart of progression of PAH treatment development. A point-of-care biopsy in PAH (pulmonary artery catheterization is required for diagnosis. It is low-risk, readily available, is often performed sequentially over time, and gives the practitioner direct access to affected tissue.

[0063] In summary, PAH is a highly morbid cardiopulmonary disease characterized by wide differences in biology (phenotype) that results in inconsistent treatment response. PAH is likely, therefore, to benefit from precision medicine, which aims to link unique molecular features with patient-specific clinical characteristics, but personalized medicine for PAH has not been successfully achieved.

[0064] We have assembled a method (computational and clinically translatable pipeline) for personalizing the trascriptomic profile of patients with pulmonary arterial hypertension (PAH), which is a highly morbid form of cardiopulmonary disease, using point-of-care acquisition of pulmonary endothelial cells. These data are then analyzed to generate a personalized protein-protein interactome that serves as the basis for a systems pharmacology approach. Quantitating the PAH drug targets in each of the personalized networks predicted performance on 6-minute walk tests following a course of therapy, as well as a number of other clinically relevant metrics in PAH, including risk assessment scores and a heart failure marker (brain natriuretic peptide levels). This analytic pipeline advances precision medicine by individualizing patient-level data acquired from effected tissue (i.e., pulmonary endothelium) and generates an evidence-based approach to drug selection. This methodological pipeline personalizes the treatment of patients with PAH and addresses a number of major obstacles in theAttorney Docket No. 15024-398PC0 approaches used in patient management and drug development, and can be applied more widely to numerous other diseases across the spectrum of medical disorders.

[0065] The invention is a method that individualizes drug treatment selection to improve clinical outcome for patients with pulmonary arterial hypertension and other similarly heterogenous diseases. In this method, a minimally invasive biopsy during routine care is used to isolate appropriate cells, such as the pulmonary artery endothelial transcriptome for PAH. Next, these data are added to a correlation matrix that includes transcript data from normal controls. Significant changes in these gene correlations that are induced by the addition of the transcriptomic data from one patient identifies gene pairs that are analyzed further. Gene pairs emerging from the correlation matrix that are significantly different after the addition of data from one patient to the matrix are mapped to the PPI. If that gene pair is found in the interactome, this indicates a functional relationship between the proteins coded by those genes. Two genes that map to the interactome form a single protein-protein interaction.

[0066] Specifically, the functionality of these gene pairs is determined by mapping them to a master resource on protein-protein interactions thereby generating individualized (patient-specific) interactomes. Alignment between the molecular target(s) of PAH drugs with the molecular features of the individualized interactome is associated with greater treatment response and, thus, used to inform the selection of a specific pharmacotherapeutic in clinical practice. This invention advances precision medicine by tailoring treatment selection based on patient-specific pathobiology and can be generalized to other complex oncologic, cardiovascular, pulmonary, autoimmune and other diseases for which affected or involved tissue sampling is possible.

[0067] Using this method, the information on molecular heterogeneity is reduced to patient-specific data that can be used to select a particular treatment modality that is most likely to be successful in that patient. The clinical disease under discussion does not matter per se. The method described here therefore can be used generally with other complex diseases defined by molecular heterogeneity that exists across patients sharing the same diagnosis. This includes other common but highly morbid and challenging to treat cardiovascular, pulmonary, connective tissue, oncological, and auto-immune diseases more broadly. Analysis of affected tissue or a direct surrogate thereof (e.g., peripheral blood, saliva, urine) is used to guide diagnosis, prognosis, and potentially treatment decision-making. This includes cardiomyopathy; parenchymal (fibrotic) lung disease; solid tumors affecting the breast, lung, colon, pancreas, ovaries; inflammatory bowel disease such as Crohn’s disease and ulcerative colitis; systemic lupus erythematosus and systemic sclerosis; and other similar disorders characterized byAttorney Docket No. 15024-398PC0 convergent endophenotypes (i.e., mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, metabolic dysfunction).

[0068] 2. Definitions

[0069] Unless defined otherwise, all technical and scientific terms use herein have the same meaning as commonly understood by one of ordinary skill in the art. Although various methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, suitable methods and materials used are described below. However, the skilled artisan understands that the methods and materials used and described are examples and may not be the only ones suitable for use in the invention. Moreover, as measurements are subject to inherent variability, any temperature, weight, volume, time interval, pH, salinity, molarity or molality, range, concentration, and any other measurements, quantities, or numerical expressions given herein are intended to be approximate and not exact or critical figures unless expressly stated to the contrary.

[0070] In the foregoing specification, the invention has been described with reference to specific embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. Throughout this specification and the claims, unless the context requires otherwise, the word “comprise” and its variations, such as “comprises” and “comprising,” will be understood to imply the inclusion of a stated item, element or step or group of items, elements or steps but not the inclusion of any other item, element, step, or group of items, elements, or steps. Furthermore, the indefinite article “a” or “an” is meant to indicate one or more of the item, element, or step modified by the article.

[0071] As used herein, the term “about” means plus or minus 20 percent of the recited value so that, for example, “about 0.125” means 0.125 ± 0.025, and “about 1.0” means 1.0 ± 0.2. Notwithstanding that the numerical ranges and parameters setting for the broad scope of the invention are approximations, the numerical values set forth in specific non-limiting examples are reported as precisely as possible. Any numerical value, however, inherently contains certain errors necessarily resulting from the standard deviation found in their respective testing measurements at the time of this writing.

[0072] As used herein, the terms “personalized or precision medicine” refers to a method that aligns the unique biological signature of a patient with the treatment that is best-suited to treat the disease in thatAttorney Docket No. 15024-398PC0 particular patient. Precision medicine aims to individualize molecular and phenotypic data, and leverage this information for optimizing clinically important parameters such as prognosis and treatment.

[0073] As used herein, the term “endophenotype” refers to a specific feature or characteristic of a disorder in a patient.

[0074] As used herein, the term “wiring map” refers to the connectivity of functional molecular pathways. The terms “patient- and time-specific wiring maps” refer to wiring maps, as defined above, that are specific to an individual patient at a particular point in time.

[0075] As used herein, the term “pulmonary vascular disease” refers to all forms of pulmonaryhypertension.

[0076] As used herein, the terms “subject” and “patient” are considered interchangeable and refer to any animal suitable for use with the method described here. The terms can include simians, humans, avians, felines, canines, equines, rodents, bovines, porcines, ovines, caprines, mammalian farm animals, mammalian sport animals, mammalian work animals, mammalian laboratory animals, and mammalian companion animals. A suitable subject for the invention preferably is a human that is suspected of having, has been diagnosed as having, or is at risk of developing a disease that can be ameliorated, modified, or treated using the method. In particular any disease that is heterogenous is preferred, for example PAH. Conditions amenable to treatment using the invention which define a appropriate subject or patient will be discerned easily by the person of skill in the art based on the disclosures herein. The term “subject in need” refers to the same group of animals, preferably humans.

[0077] As used herein, the term "administering" and its cognates refers to introducing an agent to a subject or contacting the agent with the subject, and can be performed using any of the various methods or delivery systems for administering agents, pharmaceutical compositions or delivering gene vectors known to those skilled in the art. Modes of administering include, but are not limited to oral administration or intravenous, subcutaneous, intramuscular or intraperitoneal injections, rectal administration by way of suppositories or enema, or local administration directly into or onto a target tissue, or administration by any route or method that delivers a therapeutically effective amount of the drug or composition to the cells or tissue to which it is targeted. Administration can refer to introducing the nucleic acid construct to the subject as DNA or mRNA, introducing a vector containing the nucleic acid to the subject, or introducing cells that have been transduced ex vivo with a construct, such as by electroporation or using a vector as described herein, to the subject.Attorney Docket No. 15024-398PC0

[0078] As used herein, the terms "treatment," "treating," and the like, as used herein refer to obtaining a desired pharmacologic and / or physiological effect. "Treatment," includes: (a) preventing the condition or disease or symptom thereof from occurring in a subject which may be predisposed to the condition or disease but has not yet been diagnosed as having it; (b) inhibiting the condition or disease or symptom thereof, such as, arresting its development or exacerbation; and (c) relieving, alleviating or ameliorating the condition or disease or symptom thereof, such as, for example, causing regression of the condition or disease or symptom thereof.

[0079] As used herein, the term “cell biopsy” refers to the acquisition of cells from the lung blood vessels of patients, which are used to build patient- and time-specific wiring maps.

[0080] As used herein, the phrase “characterizing the pulmonary artery endothelial cells by subtype” means cell biopsies assigned to clinical diagnostic groups of pulmonary hypertension.

[0081] As used herein, the term “transcriptomics sequencing” refers to assembling the mRNA sequence pattern of a biopsy sample using bulk RNA-seq methods.

[0082] As used herein, the phrase “pulmonary vascular disease-specific gene expression signature” refers to the transcript sequence profile that distinguishes patients with pulmonary vascular disease from other patients or healthy controls.

[0083] As used herein, the term “treatment response” refers to a clinical response induced by the administration of a treatment.

[0084] As used herein, the term “disease pathobiology” refers to the causative molecular basis of a disease.

[0085] As used herein, the term “protein-protein interactions” refers to a map of all known protein- protein interactions.

[0086] As used herein, the term “individualized protein-protein interactions” refers to a pattern or map of all known molecular interactions between proteins that is unique to an individual patient. It is based on a compilation of data from functional molecular assays such as described herein.

[0087] As used herein, the term “gene therapeutic target alignment” refers to the proportion of drug targets that are present in a protein-protein network.

[0088] As used herein, the phrase “complex disease defined by molecular heterogeneity across patients sharing the same diagnosis” refers to such diseases as the common but highly morbid and challenging to treat cardiovascular, pulmonary, connective tissue, oncological, and auto-immune diseases, for which analysis of affected tissue or a direct surrogate thereof (e.g., peripheral blood, saliva, urine) is used toAttorney Docket No. 15024-398PC0 guide diagnosis, prognosis, and treatment decision-making. This includes cardiomyopathy; parenchymal (fibrotic) lung disease; solid tumors affecting the breast, lung, colon, pancreas, ovaries; inflammatory bowel disease such as Crohn’s disease and ulcerative colitis; systemic lupus erythematosus and systemic sclerosis; and other similar disorders characterized by convergent endophenotypes (i.e., mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, metabolic dysfunction), as well as pulmonary vascular diseases such as pulmonary arterial hypertension.

[0089] 3. Summary of Results

[0090] Pulmonary arterial hypertension (PAH) is characterized by molecular and clinical heterogeneity as well as variable pharmacotherapeutic response profiles. Analytic strategies that leverage patient- specific pathogenetic signatures to guide treatment selection may advance precision medicine in PAH, but do not exist currently. We have developed a new method that generates patient-specific networks using functional or physical PPIs. As proof-of-principal, the patient-specific PPI network approach was tested in hypertrophic cardiomyopathy (HCM), which akin to PAH is a complex cardiovascular pathophenotype that involves numerous endophenotypes. The PPIs fully delineated HCM from dilated cardiomyopathy (DCM), despite a common genetic basis reported for these diseases.

[0091] We used a retrospective, blinded in silico clinical trial design to test the hypothesis that transcriptomic data from PAEC biopsies collected during routine RHC could be used to build individualized protein-protein interactomes and inform patient-level clinical response to PAH therapies. The primary exposure was the molecular target(s) of a PAH drug class prescribed at the time of RHC. The final study cohort included 25 PAH participants (56 [range: 30-84] yr; 91% female; 81%, white) and 3 controls contributing 32 and 3 cell biopsy specimens, respectively. The patient-specific PPI network approach identified 25 new HCM endophenotypes.

[0092] The patient-specific interactomes varied widely in topology (nodes median=5954 [range: 2074 – 9569]) and complexity (edges median=8494 [range: 2097 – 24877]). All 32 individualized PAH interactomes were enriched significantly with genes associated with key PAH endophenotypes such as hypoxia, oxidant stress, and apoptosis. Concordance between the molecular targets of a prescribed PAH pharmacotherapy and the topology of the corresponding individualized interactome was associated with improvement in multiple PAH metrics, including a reduction in brain natriuretic peptide levels at 6 months (β -627.6 pg / mL, [95% confidence interval (CI) -986.4, -268.8]; p = 0.003) and a decrease fromAttorney Docket No. 15024-398PC0 high to intermediate REVEAL 2.0 Risk scores at 6 months (β -1.3 units, [95% CI -2.6, -0.04]; p=0.043) that persisted at 12 months (β -1.7, [95% CI -3.4, -0.1]; p=0.059).

[0093] Therapeutic target-interactome alignment was also associated with longer six-minute walk distance and lower World Health Organization functional class at 6 and 12 months and a reduction in COMPERA 2.0 four-strata risk assessment at 6 months, from intermediate-high to low risk for mortality. There were no consistent associations identified between PPI target concordance and control comparators (diuretics and anxiolytics / antidepressants) and clinical end points. PAEC biopsy-derived patient-specific networks capture individual-level heterogeneity in PAH. Integration of transcriptomics acquired at point-of-care with network medicine may individualize treatment selection and improve clinically relevant endpoints in PAH.

[0094] Gene target-PAH therapeutic class alignment was associated with improvement in multiple important PAH end points used for drug approval or treatment escalation in clinical guidelines. Some of these relationships persisted for up to a year after introduction of therapy. The effect estimate observed for 6MWD at 6 months – over 160 m – well surpassed the minimally clinically important difference of approximately 33 m for 6MWD in PAH. Gene target-PAH therapeutic class alignment was also associated with normalization of BNP levels to low categories of PAH risk. Estimates for REVEAL 2.0 Risk and COMPERA 2.0 risk assessments were decreased by at least one risk category, which has been associated with improved survival and a reduction in important clinical events including lung transplant and hospitalizations.

[0095] In summary, the patient-specific PPI networks informed clinical variables with prognostic relevance. A validated, multicenter phenome biobank has been assembled that includes proteomic and genomic data from iPAH and HPAH patients, as well as asymptomatic relatives of PAH patients and healthy controls. In PAH, the circulating proteome is prognostic, but also highly variable across patients. These data indicate that developing patient-specific PPI networks will personalize clinical phenotyping and optimize prognosis in PAH. Findings from analyses proposed in this project will advance precision medicine by also clarifying the relationship between genetic risk and pathobiology on an individual patient level (i.e., reticulotypes), establishing a critical and much-needed framework for personalizing risk assessment and pharmacotherapeutic selection, applicable to numerous diseases.

[0096] 4. Embodiments of the Invention

[0097] A. IntroductionAttorney Docket No. 15024-398PC0

[0098] Here, we propose a unique strategy in which proteomic data from an individual PAH patient is used to perturb a control proteomic matrix, ultimately generating patient-specific PPI networks. These networks are then used to personalize the pathobiology-endophenotype relationship. The method thus takes molecular heterogeneity and reduces the information to patient-specific data that can be used for drug selection.

[0099] Data are presented which shows the feasibility of this method to classify and prognosticate patients, thereby advancing precision medicine. The aims of this method are to use PPI networks to identify molecular interactions that distinguish individual patients using averaged biological data that marginalizes certain key information. The methods were tested here with PAH but can be used in the same manner for other diseases as well, such as cardiovascular, pulmonary, connective tissue, oncological, and autoimmune diseases more broadly, for which analysis of affected tissue or a direct surrogate thereof (e.g., peripheral blood, saliva, urine) is used to guide diagnosis, prognosis, and potentially treatment decision-making. This study involved using the PAH phenome repository to establish primary and validation cohorts using a bootstrapping approach, developing the patient-specific PPI networks for the primary cohort per the methods described here, compiling PPIs that are both shared and unique among patients within and between idiopathic pulmonary arterial hypertension (IPAH) and hereditable pulmonary arterial hypertension (HPAH), and profiling static and dynamic PAH network topology over time, and in patients developing PAH during the study.

[0100] A second objective of the study was to demonstrate that patient-specific PPI networks inform clinical parameters and prognosis in PAH. This is accomplished by analyzing PPI networks enriched by endophenotype to individualize profile and prognosis. First the PPIs are defined, using pathogenic gene variants in the HPAH patient-specific networks. These data can be used to identify common but as yet unrealized functional pathways in PAH informed by gene variants. Then, a network score is developed that quantitates the PPIs enrichment for PAH the endophenotypes, and then used to inform the genetic, clinical and prognostic profile of individual patients. The network score is validated for predicting PAH clinical profile and prognosis PAH in the bootstrap cohort and the molecular inhibition / overexpression and histological methods are used to elucidate pathobiological mechanisms of key PPIs in vascular cells ex vivo and lung specimens in situ from a separate PAH cohort.

[0101] In addition, computational modeling was used to predict patient-specific drug therapies using PPIs. The drug-protein interaction databases were integrated with PPI networks to identify patient- specific pharmacotherapies. Then approved and experimental PAH drugs were validated to individualAttorney Docket No. 15024-398PC0 PPI networks, and the results verified by measuring plasma drug-target proteins levels to establish a path to rational precision-medicine clinical trial design.

[0102] B. Overall Study Design

[0103] The overall study design is presented in FIG. 3. A retrospective analysis of patients undergoing RHC for the routine clinical assessment of dyspnea or PAH monitoring at Rhode Island Hospital (Brown University Health) between 2017-2022. All patients were treated based on usual standard-of- care (informed by RHC and other results) and at the discretion of their managing PAH clinician. Clinical data were paired retrospectively with data from patient-specific PPI interactomes assembled from individual PAEC biopsies performed at the time of RHC. A systems pharmacology approach was then used to determine the molecular targets of the exposure, which in this study were FDA-approved PAH therapeutics (or non-PAH drugs as comparators) prescribed at the time of, or following, RHC. The goal was to determine if the proportion of therapeutic targets in the individual-specific network was predictive of validated clinical outcome measures assessed as part of routine clinical practice.

[0104] FIG. 3 is a study schematic for the in silico clinical trial. Clinical data was collected within six months before RHC and up to 12 months following RHC. PAH treatments were collected at the time of or after completion of the RHC, during which time a cell biopsy was obtained. Following completion of RNA sequencing, an individual PAH PPI network was constructed. The study exposure was alignment between drug class gene targets and patient specific PAH genes identified in the individual PPI. A similar approach was taken for active and inactive comparators.

[0105] C. PAH

[0106] Pulmonary arterial hypertension is a highly morbid cardiopulmonary disease characterized by an obliterative vasculopathy involving distal pulmonary arterials that promotes right heart failure. The disease adversely affects lifespan, quality of life, and healthcare expenditure. For example, the 5-year mortality rate among PAH patients is about 60%, and impaired patient health-related quality of life is similar to chronic obstructive pulmonary disease or renal failure across all parameters, including mental health burden and physical activity / function. The economic ramifications of PAH therefore are substantial: 50% of patients report total loss of employment, and annual healthcare system expenditure is $6 billion for the U.S. alone.

[0107] The treatment response to PAH drug therapies is highly variable, even under tightly controlled circumstances customary among randomized clinical trials (RCTs). Four recently completed RCTs, each widely regarded as highly successful and studying different pharmacotherapy drug classes in PAHAttorney Docket No. 15024-398PC0 are: AMBITION, SERAPHIN, GRIPHON, and PATENT-1. In these studies, a satisfactory clinical response (or equivalent measurement) was not observed in 61%, 27%, 31% and 80% of patients, respectively. Importantly, these were combination therapy studies using drugs targeting different pathways thought to be active in PAH. In ‘real world’ clinical practice, inadequate therapeutic response to guideline-based recommendations is highly prevalent. This suggests that efforts to understand variability in treatment response should include cross-sectional populations in addition to RCTs.

[0108] Molecular heterogeneity is pervasive in PAH, and a barrier to drug development and RCT design. Clinically, PAH is defined by elevated mean pulmonary artery pressure (mPAP) and pulmonary vascular resistance (PVR) without a secondary cause. However, many patients fall into a ‘grey’ area, and may change clinical diagnosis during the natural history of their disease. The clinical presentation likely is a result of numerous convergent molecular mechanisms, which differ in dominance between patients or change within a patient during the course of the illness. Many pathogenic endophenotypes (defined as a feature or characteristic of a disorder, affecting different lung vascular cell types) are implicated in PAH arterial remodeling. Specifically, apoptosis-resistance, proliferation, oxidant stress, dysregulated metabolism, fibrosis, and hypercontractility are endophenotypes described in experimental PAH in vivo and in patients in situ or clinically. However, the molecular events underlying abnormalities in PAH are integrated, and span genetic and post-transcriptional events that affect protein structure and function. Moreover, the PAH clinical spectrum is now recognized to include extra- pulmonary vascular tissues that contribute to right ventricular dysfunction (independent of increased afterload), volitional muscle atrophy, and neurohumoral dysregulation. Thus, applying unbiased analytical strategies that consider multiple molecular interactions simultaneously to understand the pathobiological profile of patients is a particularly attractive strategy for studying PAH.

[0109] In establishing the framework of precision medicine in PAH, precision medicine aims to individualize molecular and phenotypic data, and leverage this information for optimizing clinically important parameters such as prognosis and treatment. The rationale for exploring precision-medicine in PAH is based on: variable treatment responsiveness, heterogeneous pathobiology, and elevated rates of adverse outcome that persist despite the availability of 14 FDA-approved medications. In the past decade, availability of high-throughput multiplex bioassay platforms has expanded access to big datasets reporting results from genome sequencing, mRNA microarrays, and large-scale proteomics. Yet, precision medicine in PAH (and nearly all other medical disciplines) remains unrealized. Advancing precision medicine in complex cardiopulmonary diseases, including PAH, requires access to big dataAttorney Docket No. 15024-398PC0 from patient samples, analytics that emphasize functionally important signaling pathways, and innovative strategies that integrate biological and phenotypic data on an individual-patient level.

[0110] A validated resource of PAH phenomes has been established and includes whole genome sequencing as well as proteomic, clinical, and outcome data collected at two time points about 1 year apart for idiopathic (IPAH) (N=303), hereditary (HPAH) (N=41), asymptomatic family members of patients with PAH (N=48) including N=3 that developed PAH during the study, and healthy controls (N=108). We proposed to study this database to test our central hypothesis that: Developing patient- specific PPI networks will personalize clinical phenotyping and optimize prognosis in PAH. Findings from this project will also clarify the relationship between genetic risk and pathobiology on an individual patient level, establishing a rational platform for personalizing drug selection in PAH.

[0111] D. Barriers to Achieving Precision Medicine.

[0112] Barriers to achieving precision medicine include the limitations of scientific reductionism and over-reliance on averaged data. Precision-based methods for diagnosing and prognosticating PAH have focused largely on variants in BMPR222, and more recently identified putative PAH-causing genes, such as SOX17.23. This approach has yielded critical information expanding and diversifying the pathogenetic mechanisms underlying PAH. Further, progress in genetics has had a major impact clinically. This includes genetic testing and counseling in at-risk patients, which is now PAH standard of care. However, despite extensive gene sequencing of sizeable PAH populations, <30% of patients have single gene variants that cause or predispose PAH. The majority of PAH remains idiopathic. Further, cross-talk has been reported for PAH signaling pathways between pulmonary artery endothelial cells (PAECs), pulmonary artery smooth muscle cells (PASMCs), pericytes, and fibroblasts that is independent of a predisposing genetic variant per se. This indicates that a single sentinel molecular (e.g., genetic) event is unlikely to explain the totality of PAH or other similar conditions.

[0113] With respect to over-reliance on averaged data, the most common strategy to identify biological data separating patients from controls relies on comparing the mean (average) difference between groups. However, averaging biological data has two major limitations. First, important biological information from individual patients is consolidated and, thus, biologically important variation often goes overlooked. This is particularly salient for patients that contribute extreme datapoints (high- or low- values) to the cohort average. Second, transcript or protein quantity alone does not indicate functionality. Therefore, deciphering big data based on directionality or magnitude change in expression alone is often unsuccessful for determining pathogenicity.Attorney Docket No. 15024-398PC0

[0114] These barriers have created a major unmet need in the ability to PAH precision medicine. Therefore, it has not been possible previously to develop a validated strategy that both leverages knowledge on functionally important relationships between biological intermediaries and includes individual patient-level data. In addition, the strategy disclosed here also informs pathobiology- phenotype relationships, known as reticulotypes that are patient-specific and useful clinically. See FIG. 4, which provides a schematic of the general methods. FIG. 4A outlines data obtained from a population of PAH patients, biological array data such as proteomics, used to identify functionally important and patient-specific biological pathways. FIG. 4B shows that patient-specific information is analyzed further to identify PPIs that are unique to each patient and share across the population. Collectively, these data are used to tailor phenotyping, prognosis, and treatment selection. P1 = specific patient; P…N = individual patients in a population.

[0115] E. Methods of Advancing Precision Medicine

[0116] The studies described here used data from the PAH Phenome Biobank, a well-established national repository containing patient samples from the UK and France. This specification describes advanced precision medicine in PAH by using a novel network medicine-based approach that individualizes patient reticulotypes. The database includes proteomic, genomic, demographic, clinical, prognostic and outcome data from patients with idiopathic (iPAH) (N=303), hereditable (HPAH) PAH (N=41), asymptomatic relatives of PAH patients (N=48) including N=3 that developed PAH during the study period, and healthy controls (N=108). Importantly, paired samples collected about 1 year apart are available for all patients.

[0117] Classical reductionist scientific methods are useful for determining the effect of changes to a single component of a biochemical reaction. However, this strategy may oversimplify the pathogenesis of complex diseases. By contrast, network medicine emphasizes the functional and physical associations between biological components to understand human diseases. In PAH, network medicine for deciphering integrated molecular pathways in complex clinical phenotypes can identify integrated signaling pathways underlying arterial fibrosis, complement-signaling, and hypoxia-induced vascular remodeling. In these efforts, differentially expressed proteins (or the protein product of gene transcripts) were mapped to the consolidated human interactome, which is a large and validated human protein- protein (PPI) network.

[0118] For this project, we used the interactome using human physical molecular interaction data from different sources, including PPIs, protein complexes, kinase-substrate interactions, and signalingAttorney Docket No. 15024-398PC0 pathways. The PPIs were derived from several high-throughput yeast-two-hybrid studies as well as high-quality PPIs from the literature were compiled from various resources. We also collected binary PPIs from other laboratories. The current version of the interactome contains information on 15,489 proteins and 188,973 molecular interactions, which includes 11% and 24% more proteins and PPIs, respectively, compared to the prior version. In the networks, nodes represent proteins and edges represent functional or physical associations among proteins.

[0119] This approach addressed two major problems in the management of cardiopulmonary diseases with direct implications for disorders across the medical spectrum. First, with these methods, a practitioner can individualize the biological signature of a patient using point-of-care assays that involve affected tissue. Second, this approach decreases uncertainty in predicting clinical response to a therapeutic. Taken together, these increase specificity of data collected in the management of patients for informing clinical decision-making on an individual patient level and while doing so provides an evidenced-based approach to rational drug selection to improve clinical relevant disease end-point(s).

[0120] Leveraging patient-specific protein-protein interactome data form a point-of-care biopsy specimen as the substrate for systems pharmacology analyses has not been reported successfully. This approach provides a translatable and clinically relevant validation marker of our testing methods (i.e., functional capacity measured by 6-minute walk distance). The components which comprise our analytical pipeline are non-intuitive and nonobvious to experts in the field of network medicine, genomics data analysis, or clinical medicine.

[0121] F. Cell Biopsy and Other Methods

[0122] Several embodiments of the invention use a “cell biopsy” approach. Right heart catheterization (RHC) is the fundamental and required procedure to diagnose and monitor PAH and other forms of pulmonary vascular disease according to present standards. In this procedure, endothelial cells (the cell type that is dysregulated and causative for PAH) are harvested from the tips of RHC balloons. In addition, we provide a method for propagating and characterizing pulmonary artery endothelial cells (PAECs) from these balloon samples from patients with pulmonary hypertension of all subtypes and at varied stages of disease. The subtypes refer to clinical diagnostic groups of the disease, such as PAH. In another embodiment, the method includes transcriptomic sequencing on these cells and determining a PAH-specific gene expression signature that can be monitored with clinical changes to provide a real- time window into disease pathobiology.Attorney Docket No. 15024-398PC0

[0123] In other embodiments, a pipeline was created in which PAECs were harvested from patients, their transcriptomic signatures were used, and then individualized protein-protein interactions (patient- and time-specific “wiring maps”) were derived to identify gene-therapeutic target alignment. PAEC signature alignment with approved and investigational PAH treatments was associated with changes in the most widely used clinical metrics to gauge treatment response. In certain other embodiments, this PAEC harvest-in silico pipeline (or “cell biopsy”) approach stands to be positioned as the first point-of- care method to direct therapeutic screening and selection in pulmonary vascular disease.

[0124] One embodiment of this method can address widely acknowledged gaps and advance precision- based care in cardiopulmonary medicine. Specifically, the “cell biopsy” sampling method can (1) be used at the bedside by clinicians following RHC (the required diagnostic and monitoring procedure that is repeated frequently throughout a patient’s life) to inform therapeutic selection to maximize efficacy and to eliminate or avoid drugs that are unlikely to be of benefit / associated with greater side effects; and (2) be widely deployed in clinical trials for predictive and prognostic enrichment and to maximize trial conduct and efficiency (e.g., avoid enrolling subjects with no target alignment).

[0125] This approach addressed two major problems in the current management of cardiopulmonary diseases with direct implications for disorders across the medical spectrum. First, this approach individualizes the biological signature of a patient using point-of-care assay involving affected tissues. Second, this approach decreases uncertainty in predicting clinical response to a therapeutic. Taken together, these factors increase specificity of data collected in the management of patients for informing clinical decision-making on an individual patient level and while doing so provides an evidenced based approach to rational drug selection to improve clinical relevant disease end-point(s).

[0126] Thus, the method involves using patient-specific protein-protein interactome data from a point- of-care biopsy specimen as the substrate for systems pharmacology analyses. Further, one embodiment of the method provides a translatable and clinically relevant validation marker (i.e., functional capacity measured by 6-minute walk distance). These components, the analytical pipeline, are non-intuitive and non-obvious to experts in the field of network medicine, -omics data analysis, or clinical medicine. Broader data on the optimal number of control samples, and potentially the development of a standard control -omics profile, would assist in automation of the pipeline.

[0127] See FIG. 5 for a schematic showing steps taken in analysis. The cell biopsy approach involves collection of pulmonary artery endothelial cells (PAECs) from pulmonary artery (PA) catheters used during routine right heart catheterizations. PAECs are then grown in culture to confluence (up toAttorney Docket No. 15024-398PC0 passage 4), and frozen. Bulk RNAsequencing using the Illumnia HiSeq platform, and transcriptomics can then be studied and compared in PAH versus controls or among varied subtypes of pulmonary hypertension.

[0128] Table 1, below, shows the top differentially upregulated genes and ontology. These are the six significantly upregulated genes and their related ontology which corresponds to several hypothesized but less established mechanisms of PAH including oxidative stress, fatty acid oxidation, sex hormone signaling, and maladapted mitochondrial electron transport and post-translational modification. These purported pathways have not previously been demonstrated to be relevant in human PAH affected tissue during the course of the disease.

[0129] Table 1. Differentially Upregulated Genes and Ontology. Symbol Gene Name logFC Adj. P- Ontology value h

[0130] To confirm endothelial cell lineage of the PAH biopsy, see FIG. 6. This figure presents flow cytometry data of PAH biopsy. Additional PAEC markers studied were CD-31, vWF+, VE-cad+, acLDL uptake, and α-SMA-. These markers have been confirmed through passage 8. FIG. 7 presents photomicrographs of several tissues as indicated. The flow cytometry results (CD146+ / CD31+ / CD45- cells; FIG. 6) and the PAEC markers as shown confirm that the cells included in cell biopsy are in fact endothelial cells. Photomicrographs (FIG. 7A through FIG. 7E) demonstrate PAECs from commercial cells (FIG. 7A) followed by PAECs from the cell biopsy method of different patients with varied subtypes of pulmonary hypertension including a patient without PAH (FIG. 7B), idiopathic PAH (FIG. 7C), HIV associated PAH (FIG. 7D), and portopulmonary hypertension with high cardiac output (FIG.Attorney Docket No. 15024-398PC0 7E). Thus, cell biopsy captures PAECs (as indicated by the listed classical cell surface markers) from patients with varied clinical backgrounds. In order to individualize PAH pathobiology an individual PAH patients, an individual PAH patient PPI network was created as shown in the chart in FIG. 8.

[0131] The pathobiology of each individual is shown by a resulting patient-specific protein-protein interaction network for each patient. This method provides an individualized pathobiology for each patient, emphasizes functional pathways with pathobiological relevance to PAH, makes a substrate for systems pharmacology analyses, and provides a rationale for patient-specific pharmacotherapy selection for PAH.

[0132] Two control groups, with same edges preferably are used; miR: z-score, select significant miR and connect them based on the target relationship (mirTar). Average Proximity; target closeness to the individual patient network relative to the interactome (0, overlap; 1=interaction away). P value= in case of #130, it was N=141 random targets vs. N=141 drug target and assessed proximity difference; normal distribution test; Bonferonni correction: considering P value for all 5284 drugs).

[0133] In the end, assembling individualized interactomes from PAH endothelial biopsy specimens collected at point-of-care using systems pharmacology predicts treatment-response to standard-of-care pharmacotherapies. Further, data from this study (in silico clinical trial) established the framework for a precision medicine pipeline that enables individualized treatment selection in PAH patients clinically.

[0134] Greater appreciation of the clinical and molecular diversity underpinning pulmonary vascular diseases has stimulated efforts to reclassify PAH patients based on deep phenotyping. Distinct endophenotypes determined by peripheral blood signatures (i.e., inflammation, monogenic variants, others) are associated with differences in PAH severity and outcome. However, approaches used in these prior studies tended to use analytic methods that can overemphasize single molecular events or bystander (rather than functional) biomarkers to explain pathogenicity and have not been individualized or utilized to inform treatment selection successfully. Here, we used (validated) functional PPIs to assemble biopsy-specific interactomes, characterize pathobiological patterns between patients, and test PAH pharmacotherapeutic response. General agreement in network topology across patients was observed when considering proteins (nodes), but substantial variability ranging from near-total to near- absent overlap between patients (including in repeated samples from the same patients) when focusing on functional PPIs. These findings imply that phenotypic heterogeneity across patients with PAH hinges on molecular connectivity detectable in affected tissue (i.e., PAECs), rather than simple expression profiles. This has important implications for treatment in PAH, since uncoupling betweenAttorney Docket No. 15024-398PC0 transcript quantity and pathogenicity is common in complex diseases, whereas PPIs implicated in individualized interactome connectivity itself was shown in this study to be a measure of functional importance.

[0135] Despite acknowledgement of the need for deep phenotyping and precision-based treatment approaches in PAH, especially given the evolving treatment landscape, challenges remain in operationalizing these concepts. Treatment response heterogeneity has been identified in PAH clinical trials and registries based on patient comorbidities which could inform prognostic enrichment, but an equally important goal would be to enhance drug development and clinical trial efficiency via predictive enrichment (e.g., targeting patient-level molecular signatures).

[0136] At the bedside, clinician selection of precise PAH therapeutic regimens would ideally improve outcomes and health-related quality of life while minimizing side effects, burden, and cost. This proof- of-concept in silico trial – which derives individualized networks from cell biopsies of the affected tissue compartment paired to prescribed treatment and important outcomes in PAH – is the first step towards a point-of-care process for therapeutic efficacy and effectiveness screening and personalized medicine in PAH.

[0137] In summary, in this study a novel network medicine approach integrating transcriptomics from PAEC biopsies acquired from PAH patients at point-of-care was used to link individualized treatment selection with improvement in clinically relevant end points. In doing so, this work progresses precision medicine through three key advances. First, an analytical pipeline integrating functional PPIs, derived from patient cells ex vivo and modeled in silico, with systems pharmacology was developed to anticipate patient-level therapeutic responses. Second, the potential for individualized interactomes to harbor clinically actionable data was demonstrated. This was tested with a clinical research design pairing patient-specific pathobiological readouts to drug effects. Third, findings from this study emphasize the broad translational potential of cell biopsies collected during routine clinical practice despite (or perhaps as a function of) variability across samples. The findings presented here provide a basis for scaling patient-specific clinical treatment responses to standard-of-care therapy in PAH, with implications for other complex diseases characterized by both diverse pathogenetic causes and variable outcomes.

[0138] Alternative conditions for which the technology can be used includes cardiomyopathy; parenchymal (fibrotic) lung diseases including idiopathic pulmonary fibrosis, sarcoidosis, hypersensitivity pneumonitis and interstitial lung diseases; solid tumors affecting the breast, lung, colon, pancreas, ovaries; inflammatory bowel disease such as Crohn’s disease and ulcerative colitis; mixedAttorney Docket No. 15024-398PC0 connective tissue disease; systemic lupus erythematosus and systemic sclerosis; and other similar disorders characterized by convergent endophenotypes (i.e., mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, metabolic dysfunction). These instances are proposed based on their biological heterogeneity, need for individualized therapy for patients, and access to affected tissue in routine clinical practice.

[0139] Cell biopsy samples are chosen to include the cells that are affected by the disease in question. Therefore, a person of skill in the art is aware of the cells or cell surrogates that should be sampled for any such disease. As an example, left ventricle myocardial cells are collected for using the inventive method for hypertrophic cardiomyopathy (HCM) and pulmonary endothelial cells are collected for using the inventive method for PAH.

[0140] 5. Examples

[0141] This invention is not limited to the particular processes, compounds, compositions, or methods described, as these may vary. The terminology used in the description is for the purpose of describing the particular versions or embodiments only, and is not intended to limit the scope of the present invention which will be limited only by the appended claims. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of embodiments of the present invention, the preferred methods, devices, systems, compounds, compositions and materials are now described.

[0142] Example 1: General Methods and Materials.

[0143] A. Clinical Study Design and Data Collection

[0144] Participants were included in this study if they had a clinical diagnosis of PAH using the World Symposium on Pulmonary Hypertension criteria consistent with this diagnosis (confirmed post hoc), had a successful PAEC biopsy from RHC, and had ≥1 follow-up clinical assessment after RHC. Baseline clinical data were collected for participants ≤6 months before the date of the study RHC. Post-RHC clinical assessments (e.g., 6-minute walk distance [6MWD], risk scores) that occurred up to 12 months after initiation of PAH therapies were included as study end points. See Table 2, below, for a flow chart showing the steps involved in performing the invention. Suitable cells are isolated, preferably through a minimally invasive technique, performed at a convenient time such as at the time of right heartAttorney Docket No. 15024-398PC0 catheterization which the disease is PAH. The cells are expanded and mRNA is isolated from the patients and from control subjects.

[0145] Table 2. Method Flow Chart. Step 1 Pulmonary endothelial cells from patients with pulmonary arterial dAttorney Docket No. 15024-398PC0

[0146] B. Clinical End Points

[0147] Widely employed and established clinical metrics in PAH were selected as study end points. The six-minute walk test is a standardized, self-paced test of exercise capacity which captures 6MWD. The 6MWD is the primary end point for almost all Phase III trials and the basis for drug approval in PAH. The minimal clinically important distance for 6MWD is 30-35 m. Brain-natriuretic peptide (BNP) and N-terminal pro-BNP (NT-proBNP) levels are non-specific markers of right ventricular stretch that are prognostically meaningful in PAH. The cut-off values for BNP of <50, 50-199, 200-800, and >800 ng / L are ascribed to escalating risk in various multiparameter prognostic models. Several of these risk assessment scores have been developed for the prognosis and monitoring of PAH and are recommended at diagnosis and frequently thereafter. World Health Organization functional class (WHO-FC), 6MWD, and BNP / NT-proBNP are the most informative components in all scores. Functional class is one of the most established clinical parameters evaluated in PAH and until recently was the basis of therapeutic escalation.

[0148] The United States Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) 2.0 risk calculator has been validated as a prognostic score for predicting 1-year survival. REVEAL 2.0 incorporates 13 total items including 6MWD, functional class and BNP as well as PAH subgroup, demographic information, hospitalizations, echocardiographic, vital sign, hemodynamic, pulmonary and renal function data. Seven components are required to generate a score and missing items are assigned “0”s otherwise; risk scores of 0-6, 7-8, and ≥9 correspond to low, intermediate and high-risk for mortality. The Comparative, Prospective Registry of Newly Initiated Therapies for PH (COMPERA) 2.0 four-strata risk assessment (1=low; 2=intermediate-low; 3=intermediate high; 4=high) for mortality incorporates functional class, 6MWD, and BNP / NT-proBNP. COMPERA 2.0 four-strata risk assessment is recommended to follow PAH patients longitudinally and is the basis for treatment escalation in the most recent consensus guidelines. Both the clinical data extraction and bioinformatic team members worked independently such that PAH treatments and clinical metrics as well as PPI network results were blinded, respectively.

[0149] C. Systems Pharmacology

[0150] Ten Food and Drug Administration-approved drugs for PAH, five calcium channel blockers and vasodilators that were previously included in the PAH module and the investigational drugs DHEA (dehydroepiandrosterone) and sotatercept were considered and given to participants in our center and enrolled in these trials. See Tables 3 and 4, below.Attorney Docket No. 15024-398PC0

[0151] Table 3. Primary clinical indication and treatment condition for pharmacotherapies actively prescribed to participants in the current study. Pharmacotherapy Primary Clinical Treatment Indication Conditiont PAH and no overlap in the PAH module. PAH=pulmonary arterial hypertension.

[0152] Table 4. Drugs Analyzed in the Study and Their Corresponding DrugBank Molecular Targets Drug Drugbank TargetsAttorney Docket No. 15024-398PC0 Selexipag PTGIR Diltiazem CACNG1 A, , , ,

[0153] D. Control Comparators

[0154] Serotonin synthesis and signaling is implicated in PAH pathobiology. Many observational studies have demonstrated relationships between antidepressants, specifically selective serotoninAttorney Docket No. 15024-398PC0 reuptake inhibitors, and risk in variable forms of pulmonary vascular disease. More recently, rodatristat ethyl was developed to enhance serotonin synthesis by peripheral inhibition of tryptophan hydroxylase 1 (TPH1) and tested in a Phase 2b study terminated for safety. Similar relationships were observed in the current study, where psychotropic medications targeting serotonin signaling were loosely associated with worse PAH metrics. Confirmation of overlap of these drugs in the PAH module, detection of related gene targets in our cell biopsy samples, and linking of greater target engagement with PAH outcomes in a direction that was predicted by empiric research provides key measure of validity of to the approach described here as well as insight into molecular phenotypes in PAH patients per se that may be informative in clinical studies involving these drugs.

[0155] There were no relationships detected between PPI alignment and diuretics outside of the PAH module (a pure “placebo” condition) and PAH end points, as hypothesized given no direct molecular targets were found (nor anticipated) in the PAH module. Target alignment with non-serotonin psychotropics were associated with improved REVEAL 2.0 Risk scores, but had no associations with 6MWD, BNP or functional class. As these treatments had no overlap in the PAH module (and therefore would be unlikely to target biologic pathways relevant to pulmonary vascular disease in the study participants), we speculate this could be due to confounding by indication. In other words, patients who were experiencing progressive disease (i.e., high REVEAL 2.0 Risk scores at the time of cell biopsy) were more likely to be prescribed psychotropic medications, given the known high concurrent rate of anxiety / depression with PAH, with subsequent reduction in REVEAL scores over time with PAH treatment.

[0156] The targets of the drugs (from DrugBank and seen in Table 3, above) were used. It was calculated whether the targets of the drugs were in the individualized PPI networks. If so, the number of targets (among all the targets of the drugs) were quantitated. Target-interactome alignment, the exposure for this study, was defined as the proportion of protein targets for a given PAH drug class (e.g., phosphodiesterase-5 inhibitors) in a cell biopsy PPI network out of all the possible protein targets in that same drug class (e.g., phosphodiesterase-5 inhibitors). In other words, target-interactome alignment was the proportion of drug class targets present in the cell-biopsy-derived individualized PPI interactome for the therapy administered to a corresponding patient.

[0157] To simulate a placebo-controlled condition in this in silico clinical study, we chose concomitant medications that are prescribed frequently in PAH but are not considered PAH-targeted therapies nor are they FDA-approved for the treatment of PAH. We assessed whether the proportion of protein targets forAttorney Docket No. 15024-398PC0 those concomitant drug classes in the sample-specific network was associated with the same clinical end points described above. Specifically, the protein targets of individual diuretic, antidepressant, and anxiolytic medications were compiled from DrugBank. See Tables 2 and 3, above. The degree of overlap of the drug targets of these ‘placebo’ drugs to the PAH disease module was calculated to make sure that the ‘placebo’ drugs were not significantly close to the PAH disease module (in essence, the degree to which inactive comparators could be considered active comparators).

[0158] Diuretics, antidepressants and anxiolytics were separated into two groups: those with PAH module overlap and those without. Spironolactone was the only diuretic with overlap among the prescribed diuretics. Based on mechanism of action and correlative experimental data, several antidepressants targeting serotonin signaling (venlafaxine, fluoxetine, quetiapine, escitalopram, desvenlafaxine, mirtazapine, citalopram) had potential overlap with the PAH module and were, therefore, analyzed separately (i.e., active comparators; grouped as Serotonin Psychotropics) from those that did not (bupropion and benzodiazepines including clonazepam, alprazolam, lorazepam) (i.e., inactive comparators; grouped as Non-Serotonin Psychotropics).

[0159] E. Isolation and Propagation of Pulmonary Artery Endothelial Cells (PAECs) During RightHeart Catheterization (RHC).

[0160] Detailed methods and confirmation of PAEC lineage have been published previously. In brief, following RHC, the pulmonary artery catheter was retracted into the catheter sheath and both were removed en bloc from the patient. The catheter tip was advanced out of the sheath and placed directly into warm media (37ºC) (EndoGRO; Millipore SigmaTM) with the balloon deflated and immediately into a heater for travel to the laboratory. The catheter tip with balloon was placed directly into one well of a 24-well plate with Attachment Factor Solution (Cell ApplicationsTM, Inc) and washed with fresh media (EndoGRO-vascular endothelial growth factor (VEGF) complete media kit; Millipore SigmaTM) every two days. Cells were then seeded into T-25 and T-75 flasks until passaged cells reached confluence over the next four to five days. Only the clinician conducting the RHC was aware of patient characteristics. The remainder of the study staff were blinded.

[0161] F. RNA Sequencing and Analysis

[0162] Transcriptomic analyses were performed on mRNA isolated from PAECs at cell passage 3 or 4 for all samples; we had previously demonstrated transcriptome stability across these passages. Library preparation and RNA sequencing was conducted through AzentaTM. The same RNA isolation and processing technique was used for all specimens. Specimens were submitted for sequencing in batchesAttorney Docket No. 15024-398PC0 and grouped randomly within each batch. Libraries were sequenced using a 2 × 50 bp paired end rapid run on the IlluminaTMHiSeq2500 platform. Each sample batch was sequenced all at once and distributed equally across sequencing lanes. Technical replicates were not included. All specimens had RIN values > 6 and DV200 > 70%. FastQ files were aligned to the GRCh38 reference sequence using HISAT2 and gene IDs were resolved using annotateMyIDs.

[0163] Across the entire study cohort, there were three batches of RNA sequencing data. The first two batches included 25,702 genes from 10 unique PAH samples (from 9 PAH participants) and 3 unique non-disease controls (individuals referred for RHC for the evaluation of dyspnea but without pulmonary hypertension), and 28,395 genes from 12 individual PAH samples (from 9 PAH participants), respectively. The third batch included 10 additional PAH samples (from 10 PAH participants). The three datasets were merged together and any transcript with <25 read counts in all the samples was removed to ensure the genes were expressed in either non-disease controls or PAH samples. The final dataset included 32 PAH samples (from 25 unique PAH participants), 3 non-disease controls, and 12,562 genes.

[0164] G. Statistical Analysis

[0165] Summary data are presented as median (range) and categorical data as frequency (percentage). Outcomes were modeled as a function of the proportion of protein targets in the sample-specific networks by drug class. Generalized mixed modeling with sandwich estimation assuming normal distributions with the GLIMMIX procedure was employed to account for repeated samples from some participants (25 subjects had a total of 7 biological replicates; 1 subject with 5 balloon tips, 3 subjects with 2 balloon tips each) and time. Interval estimates were calculated for 95% confidence and alpha was established at the 0.05 level. Analyses were conducting using SAS Software 9.4 (SAS Inc. Cary, NC).

[0166] Example 2: Study Design.

[0167] Study participants were divided into two groups: group A (treatment naïve) and group B (on background therapy). See FIG. 9. In the study, the endpoints follow the introduction of treatment and both the treatment start and endpoints come after the interactome is sampled. In these studies, 6-minute walk distance (i.e., the distance in meters ambulated uncoached on a flat surface during a 6 minute period) was assessed per American Thoracic Society guidelines, World Health Organization Functional Class was assessed based on response to standardized question, brain natriuretic peptide level, which is a biochemical marker of heart failure, was assessed by peripheral blood measurement, and the REVEALAttorney Docket No. 15024-398PC0 2.0 Lite score was used to assess clinical risk. These indices are standard in PAH for determining disease severity and assessing clinical response to treatment. Results of the bioinformatics analysis is provided in FIG. 10 and FIG. 11. N = 3 (controls); N = 22 (PAH). Adjusted p < 0.05; (detal_PCC) >0.5. FIG. 11A shows node overlap similarity among the individual PAH networks; FIG. 11B shows edge overlap similarity among the individual PAH networks.

[0168] Example 3: Cohort Characteristics.

[0169] The final study cohort included 25 PAH participants and 3 controls (individuals who underwent RHC without pulmonary hypertension) contributing 32 and 3 cell biopsy specimens, respectively. The average age of PAH participants was 56 (range 30 – 84) years old and most were female (91%) and white (81%). Two participants (6%) were on no PAH-specific treatment at the time of RHC. Most participants were WHO-FC II at the time of RHC with a median 6MWD of 375 (range 150 – 575) m and a median REVEAL 2.0 Risk score of 10 (range 4 – 12). Table 4, below, includes characteristics of the participants at the time of RHC.

[0170] In Table 5, data are represented as n (%) or median (range). *25 subjects had a total of 7 biological replicates (1 subject with 5 balloon tips, 3 subjects with 2 balloon tips). Exercise-induced pulmonary hypertension defined as resting mPAP < 20 mmHg and with exercise, ΔmPAP / ΔCO (by thermodilution) > 3 mmHg / L / min. PAH=pulmonary arterial hypertension, defined as mean pulmonary artery pressure > 20 mmHg, pulmonary artery occlusion pressure < 15 mmHg, and pulmonary vascular resistance > 3.0 Woods units. PAOP=pulmonary artery occlusion pressure; CTD=connective tissue disease; HIV=human immunodeficiency virus; APAH=associated pulmonary arterial hypertension; PDE5=phosphodiesterase type 5.

[0171] Table 5. Baseline characteristics of subjects, by endothelial cell samples PAH ControlsAttorney Docket No. 15024-398PC0 White 26 (81) 2 (67) Black 2 (6) .Attorney Docket No. 15024-398PC0

[0172] With respect to the transcriptomic profile of participants, the PAEC RNA-Seq transcriptomic dataset, after filtering, included 32 PAH samples (from 25 unique PAH patients), 3 non-disease control samples, and 12,562 genes. The multidimensional scaling plot of the dataset shows that the variance in gene expression explained by the first and second principal component vectors are 30% and 9%, respectively, with separation between the non-disease controls and PAH groups, but also substantial separation within the PAH group. See FIG. 12. The coefficient of variation (%CV) range for the expression of genes among PAH samples was 33–545%, with 6,545 (52%) genes having a %CV ≥80%. See FIG. 13. The average correlation coefficient for replicates in non-disease controls is 0.91, whereas the average correlation coefficient of replicates in PAH samples is 0.81. See FIG. 14. These collective data suggest important pathobiological variability across the PAH samples and are consistent with the clinical heterogeneity observed in this (and many other) study populations.

[0173] Example 4: PAH Endothelial Cell Biopsy-Derived Individualized PPI Networks.

[0174] For the construction of individualized PPI networks, sample-specific networks were constructed by starting with a gene correlation matrix assembled from the non-disease control samples. The Pearson correlation coefficient (PCC) was calculated for each gene pair across all control samples (N=3). Then, a single PAH sample was added and the PCCs were recalculated. If the PCC of a gene pair was significantly changed (i.e., ΔPCC was significant) according to a pre-defined statistical distribution we the gene pair was retained. This calculation was repeated for each gene pair. After Bonferroni correction for multiple comparisons and enforcing a threshold on ΔPCC (ΔPCC>0.5), significant gene pairs were mapped to the human protein-protein interactome (PPI). The human interactome, consolidated from major high-throughput studies, consists of 17,350 proteins (gene products) and 350,950 experimental physical protein-protein interactions. The gene pairs that had significant ΔPCC and had functional protein-protein interactions in the human interactome form the specific network for a given PAH sample / patient. This was repeated for each sample to result in a sample-specific network for each cell biopsy.

[0175] The individualized PAH PPI networks varied widely in topology (median=5954 [2074 min, 9569 max] number of nodes) and complexity (median=8494 [2097 min, 24877 max] number of edges). See Table 6, Table 7 and FIG. 15. FIG. 15 is an illustration of some patient-specific networks, described in Table 7. Select clinical data is presented in Table 5 by endothelial cell sample and after adjustment for multiple comparisions. PAH=pulmonary arterial hypertension. PVR=pulmonary vascular resistance.Attorney Docket No. 15024-398PC0 WU=Wood units. HIV=human immunodeficiency virus. For Table 6, participant sample numbers represent individual patients (PXX). Letter designations represent the same patients with repeated endothelial cell samples (PXXA, B, etc). EC=endothelial cell; PCC=Pearson’s correlation coefficient.

[0176] We used the overlap coefficient, a similarity measure that measures the overlap between two finite sets, to evaluate the node similarity and edge similarity of two individualized PAH networks. The pair-wise node similarity of individualized networks displays a modular pattern, with an average 0.77. The edge similarity of individualized networks varies substantially, from 0.12 to 0.57. See FIG. 16.

[0177] Table 6. Protein-protein Interaction Network Details Adjusted p<0.05, |delta PCC|>0.5 Partici ant* EC Sam le # of roteins # of interactionsAttorney Docket No. 15024-398PC0 P21 EC27 7254 12650 P22 EC288828 20023

[0178] Table 7. Description of Specimens (FIG. 15) Specimen Age ID (years) Sex PAH Subtype PVR (WU)

[0179] A much greater overlap for nodes was observed compared with edges across the PAH networks. The uniqueness of sample-specific networks (i.e., the nodes and edges that do not appear in any other networks) varied from 0 to 37 nodes (median=3), and from 8 to 4119 edges (median=219). This finding emphasizes that it is the wiring diagrams (interactions), rather than the proteins themselves that distinguished patients in this study. Topological analyses of the individualized PAH PPI networks show that, while all the networks have similar density, they have a wide range of network diameter, characteristic path length, and size of largest connected components. See Table 8, below.

[0180] Table 8. Topological properties of individualized PAH PPI networks. Con- Size of Net- Charac Average Net- k o- ty 8 0 2 3 4 1Attorney Docket No. 15024-398PC0 P008 7654 14228 86 7455 16 6.128 3.786 0.012 0.001 1.331 P010A 7347 12724 91 7135 18 6.315 3.532 0.018 0.001 1.376 9 1 2 3 3 3 7 9 5 0 4 0 6 2 0 8 9 1 5 7 7 4 7 1

[0181] For topological analysis and endophenotype enrichment of sample-specific networks, the topological analysis of the individualized PAH networks shows that while all the networks have similar density, centralization, and heterogeneity, larger networks tend to have a smaller number of connected components, larger size of the largest connected component, and shorter diameters and characteristic path length (see Table 8). The principal clinical features of PAH are established. We collected genes associated with 75 representative endophenotypes that might be relevant to PAH and also have more than 20 associated genes from Phenopedia, DisGeNet, and OpenTargets. Since hypoxia, oxidative stress, angiogenesis, cell proliferation, and immune response are common endophenotypes in cardiovascular diseases, these were added to the endophenotype list. We compiled hypoxia-responsive genes from our in-house microarray gene expression data and a public microarray dataset. Genes associated with oxidative stress, angiogenesis, cell proliferation, apoptosis, immune response wereAttorney Docket No. 15024-398PC0 extracted from the Gene Ontology (GO) database (http: / / geneontology.org / ) by using the corresponding GO terms. In total, 80 endophenotypes were considered. We next used the hypergeometric test to determine whether an individualized PAH network is significantly enriched with genes associated with these endophenotypes compared to the human interactome as the background.

[0182] Table 9. Quantitative rank order of the distribution of PAH molecular targets in the individualized networks. Total number of Participant Molecular targets PAH targets in ID code , , 1, , AAttorney Docket No. 15024-398PC0 EDNRA, EDNRB, PTGIR, PDE4D, PLAT, PTGIR, HRH1, ADRA1B, P012A CALM1, CACNA2D1, PPARA, AR, GRIN3B, GRIN2D, GABRE, SIGMAR1 ,Attorney Docket No. 15024-398PC0 EDNRA, PLAT, PPARD, CALM1, CACNB2, CACNA1H, PDE1A, AR, 12 P25 GABRE, HSD17B1, SIGMAR1, ACVR2Ainflammation, levels of oxidative stress, hypertrophy, etc. To demonstrate that our patient-specific networks can capture individual-level pathobiological differences and phenotypic heterogeneity, we conducted an endophenotype enrichment analysis. Out of 75 endophenotypes, 34 endophenotypes were significantly (p < 0.05 adjusted by the Benjamini-Hochberg procedure) enriched in at least one sample. A heatmap of -log10P is shown in FIG. 17. All of the individualized PAH interactomes were significantly enriched with genes associated with hypoxia, oxidative stress, apoptosis and the majority of the individualized PAH interactomes were significantly enriched with DNA damage, angiogenesis, cell proliferation and immune response. There was greater variation across PAH patients in enrichment for genes associated with fibrosis, thrombosis, and inflammation. The results demonstrate the phenotypic heterogeneity of these patients and the capability of patient-specific networks to capture individual-level differences. See FIG. 17. In this figure, the rows correspond to significant endophenotypes, and the columns denote the cell biopsies from individual participants. The heatmap is depicted as -log10P, where P is the P-value from the endophenotype enrichment analysis after Bonferroni correction.

[0184] Example 5: PAH Therapeutic Target-interactome Alignment and PAH Endpoints.

[0185] We identified a PAH, diuretic, serotonin psychotropic and non-serotonin psychotropic pharmacotherapeutic target in 31, 3, 9 and 2 individualized interactomes, respectively. A heatmap of proportion of targets within a drug class expressed by each participant is provide in FIG. 18A. The quantitative rank-order summary of patient-specific PAH and comparator pharmacotherapeutic molecular targets is provided in Table 9, Table 10, and Table 11. The data in these tables indicate a non-normal distribution of targets in individualized networks for either PAH or comparator drugs. Notably, there was wide variation in the proportion of individuals expressing targets for PAH-specific classes. See FIG. 18B.

[0186] FIG. 18 shows the heterogeneity of patient-specific PAH pharmacotherapeutic molecular targets expressed in cell biopsy samples. FIG. 18A is a heatmap of proportion of targets within a drug class expressed by each participant. Row = individual participant. Column = drug class. Cell = proportion of targets expressed by an individual patient. Purple = 100% targets expressed. Peach = 0% targetsAttorney Docket No. 15024-398PC0 expressed. sGC = stimulator of soluble guanylate cyclase. DHEA = dehydroepiandrosterone. FIG. 18B shows the proportion of participants who express each target. CCB = calcium channel blocker. PDE5i = phosphodiesterase 5 inhibitor. sGC = stimulator of soluble guanylate cyclase. ERA = endothelin receptor antagonist. DHEA = dehydroepiandrosterone.

[0187] Table 10. Quantitative rank order of the distribution of PAH molecular targets in the individualized networks, categorized by drug class.Attorney Docket No. 15024-398PC0 Total number Molecular Targets of PAH 1, 1, 1 1, 1 , , 1, , 1 1, 1 ,Attorney Docket No. 15024-398PC0 GABRE, SIGMAR1 , 1 , 1 , 1, 1 , 1, , 1 , 1 1, 1Attorney Docket No. 15024-398PC0 PLAT, GRIN2D, PPARD HSD17B1, 1 , , 1 , 1 1 1 , 1, 1Attorney Docket No. 15024-398PC0 PPARA, AR, CACNA2D1, PDE4D, 1, 1 1, 1 1 1 , 1, 1 , 1, 1Attorney Docket No. 15024-398PC0 CACNA1H, HSD17B1, PDE1A SIGMAR1 ;

[0188] Table 11. Quantitative rank order of the distribution of diuretic comparators molecular targets in the individualized networks.Attorney Docket No. 15024-398PC0 Total number of placebo targets in Participant ID

[0189] Table 12. Quantitative rank order of the distribution of serotonergic psychotropic comparators molecular targets in the individualized networks.Attorney Docket No. 15024-398PC0 Total number of Placebo targets Participant SerotonergicAttorney Docket No. 15024-398PC0

[0190] Table 13. Quantitative rank order of the distribution of non-serotonergic psychotropic comparators molecular targets in the individualized networks. No. placebo targets in Participant Non-Serotonergic corresp. individualized ID code Molecular targetsAttorney Docket No. 15024-398PC0

[0191] Tables 10 through 13 indicates a greater within-class therapeutic target-interactome alignment was associated with improvements in multiple PAH end points. For example, complete concordance was associated with over a ~160 m increase at 6 months and a ~55 m increase in 6MWD at 12 months, although the overall associations did not reach statistical significance likely due to sample size (e.g., 6 month β 168 meters [95% confidence interval (CI) 106.1, 442.3], p = 0.204)). See FIG. 19A and FIG. 19B. Similarly, we observed a transition in BNP levels from elevated at baseline to within the normal range that corresponded to an absolute decrease of about 627 pg / mL (β -627.6, 95% CI -986.4, -268.8]; p = 0.003) at 6 months (FIG. 19C), an association which persisted but was not significant at 12 months (β -129.6, 95% CI -338.8, 79.6]; p = 0.22) (FIG. 19D). Therapeutic class target-interactome alignment was also likely associated with WHO FC improvement at 6 (p = 0.11) and 12 months (p = 0.063) (FIG. 19E and FIG. 19F. FIG. 19 presents data on the association between PAH therapeutic target- interactome alignment and six-minute walk distance (6MWD) (FIG. 19A and FIG. 19B), brain natriuretic peptide (BNP) levels (FIG. 19C and FIG. 19D) and functional class (FIG. 19E and FIG. 19F) at six and twelve months after right heart catheterization (RHC).

[0192] Greater therapeutic target-interactome alignment was associated with a significant improvement in REVEAL 2.0 Risk score at 6 months (β -1.3, [95% CI -2.6, -0.04]; p=0.043) (see FIG. 20) and 12 months (β -1.7, [95% CI -3.4, 0.07]; p=0.059)(see FIG. 20B), generally spanning from intermediate (7 – 8) to high risk (≥ 9) for no alignment to intermediate-low risk (≤ 8) for mortality with complete alignment. There may have also been an association between greater therapeutic class target-treatment concordance and lower COMPERA 2.0 four-strata risk assessments at 6 months (β -2.5 [95% CI -16.4, 11.3]; p = 0.26)(see FIG. 20C), an effect not seen at 12 months (see FIG. 20D).

[0193] FIG. 20A through FIG. 20D show the association between PAH therapeutic target-interactome alignment and improvement in the United States Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) 2.0 risk score (see FIG. 20A and FIG. 20B) and the Comparative, Prospective Registry of Newly Initiated Therapies for PH (COMPERA) 2.0 four-strata risk assessment (see FIG. 20C and FIG. 20D) at six and twelve months after right heart catheterization (RHC), respectively.

[0194] Example 6: Temporal Change of the Patient-specific Networks over Different Samples.

[0195] Select clinical data is presented in the Table 13, below. See also FIG. 21 (PAH=pulmonary arterial hypertension. PVR=pulmonary vascular resistance. WU=Wood units).Attorney Docket No. 15024-398PC0

[0196] Table 14. Clinical Data. Time Between Specimen Biopsies Age )

[0197] Temporal changes in the individualized networks over the disease course were examined for the same participants over different cell biopsies (4 PAH patients had repeated RHC over the course of the study for clinical indications (P005 had 5 biopsies, and P0008, P010, P012 had 2 biopsies, each). For example, the networks of P008A and P008B exhibit significant changes. While near 50% of the nodes coincide, the wiring diagram differs considerably, with only 1,092 common edges. A similar pattern was observed in participant P012. In contrast, the patient-specific network for P010 shows only moderate changes. The two networks have significant overlap, sharing 6,039 proteins and 3,802 interactions. See FIG. 21.

[0198] While P005A, P005B, P005C have similar number of nodes and edges, P005D is larger and P005E is largest. There are 827 proteins and 27 interactions common to all these five sample-specific networks. Among PAH patients with repeated samples, the proportion of targets within a drug class expressed varied within and across patients. See FIG. 22.

[0199] FIG. 23 presents heatmaps of proportion of targets within a drug class expressed by participants with biological replicates. FIG. 23A: A 56-year-old woman with portopulmonary hypertension had progressive disease with escalating therapy over serial RHCs (P005A-D, pulmonary vascular resistance (PVR) 1.7 wood units (WU), 4.0 WU, 3.7 WU, and 3.5 WU respectively). A final sample was obtained after liver transplant at which time the patient was no longer taking pulmonary vasodilators and hemodynamics had normalized (P005E, PVR 4.0 WU). FIG. 23B: A 42-year-old-woman with a known diagnosis of idiopathic PAH on PDE5i and ERA experienced persistent dyspnea and right ventricularAttorney Docket No. 15024-398PC0 dysfunction on echocardiogram (P008A, PVR 4.8 WU). She was enrolled in STELLAR, a phase 3, multicenter, double-blind, randomized, placebo-controlled trial of sotatercept in which the patient was randomized to the treatment arm (NCT 04576988). A second RHC was performed as part of the study at which time she reported improved dyspnea (P008B, 7.7 WU).

[0200] FIG. 23C: A 44-year-old man with systemic sclerosis-associated PAH experienced clinical worsening (P010B, PVR 8.9 WU) ultimately requiring lung transplant. Pathologic examination of the explant demonstrated pulmonary venule occlusion. FIG. 23D: A 68-year-old woman with limited systemic sclerosis and progressive dyspnea was diagnosed with PAH (P012A, 3.5 WU) and had progressive dyspnea requiring a follow up RHC (P012B, PVR 8.9 WU).

[0201] Example 7: Proof of Principle Data Summary.

[0202] A. The UK and French PAH Phenome Biobank Collaborative

[0203] In the UK, all patients with suspected PAH attend one of 8 designated centers for diagnosis and management decisions. Invasive lung vascular biopsy is not technically feasible in PAH at present, however, since 2013 these centers have participated in the UK National PAH Cohort study. In international collaborative efforts, samples from the National PAH Program in Paris, France are also available for use in this project. The current proposal focuses on the circulating plasma proteome. Specifically, a program has led a major initiative prospectively analyzing N=5,000 circulating proteins in the plasma of N=344 consecutive patients with iPAH (N=303), HPAH (N=41), asymptomatic family members of patients with PAH (N=48) or healthy controls (N=108) using the SomaScanV4 SomaLogicTMplatform (SomalogicTMInc.).

[0204] This platform is a valid, aptamer-based assay with excellent precision in plasma (median total % coefficient of variance [CV]=4.6%). Biological information captured for patients in this repository also includes whole genome sequence data, which has identified many novel PAH genetic variants, including ATP13A3, AQP1, GDF2 and SOX17. Audited clinical phenotype data (up to 200 fields per patient) are collected on a web-based program (OpenClinicaTM) in the UK, or via other automated and validated tools in France.

[0205] B. Sample Collection and Preparation.

[0206] Patients were sampled in a non-fasting state during routine clinical appointment visits by peripheral venous blood phlebotomy using EDTA, and stored immediately at -80℃ prior to analysis by the SomaLogicTMV3 platform. The diagnostic criteria for IPAH and HPAP were according to theAttorney Docket No. 15024-398PC0 accepted definition during the study period: mPAP >25 mm Hg, with pulmonary artery wedge pressure <15 mm Hg, and PVR >3 WU at rest with exclusion of known associated diseases. Healthy controls were collected from centers across the UK for comparison.

[0207] C. The Plasma Proteome is Prognostic in PAH.

[0208] In 2017, Rhodes and colleagues identified 20 proteins differentiating survivor vs. non-survivors in 143 consecutive iPAH or HPAH patients using the aforementioned SomaLogic aptamer screening tool. This cohort represents a large proportion of the cohort in the current proposal. Rhodes et al. showed that an unbiased quantitative approach could identify nine proteins that were both prognostic independent of plasma NT-pro-BNP concentration, the standard prognostic biomarker in PAH, and highly predictive of adverse clinical outcome. The addition of the protein panel to the REVEAL risk score, a standardized clinical tool for determining prognosis at point-of-care, increases the area under the curve from 0.83 to 0.91 (p<0.0001). In that study, clinically important proteins were implicated in endothelial dysfunction, iron bioactivity, and coagulation among other PAH endophenotypes. In the directed approach, proteins are selected for analysis based on mechanistic or other empiric data. Like the unbiased approach, this strategy has also yielded information identifying proteins that are informative clinically, including Grelmin-157, and NEDD9. Our group has shown that plasma NEDD9 is increased by 1.82-fold (P<0.0001) in a cohort of 239 PAH patients compared to controls (N=39), corresponding to a 75% increase in lung-transplant survival (95% CI, 1.12-2.73, P=0.01) per 5 ng / mL increase.

[0209] D. Limitations to Current Proteomic Phenotyping Methods.

[0210] These findings demonstrate that the circulating proteomic signature is important clinically in PAH. However, both unbiased and directed approaches for analyzing plasma proteins in PAH are associated with important limitations. In the unbiased approach, the functional significance of individual proteins or functionally significant relationships between proteins (i.e., PPIs) are not known. This, in turn, hinders the potential for individualizing therapy based on proteomic data and limits precision medicine. The directed approach relies largely on reductionism and, thus, is likely to overemphasize the importance of a single biological intermediary. These collective limitations set the framework for applying our novel patient-specific network medicine approach to phenotyping and prognosticating PAH patients, which begins with a proteomic screen, but leverages functional PPIs to discovery patient-specific reticulotypes.Attorney Docket No. 15024-398PC0

[0211] Example 8: Comparator target-interactome alignment and PAH endpoints.

[0212] Among the 32 participants in this study, 20 (80%), 8 (32%) and 7 (28%) were prescribed diuretics, or drugs included in the Serotonin Psychotropic and Non-Serotonin Psychotropic groups, respectively, at the time of RHC. Spironolactone was the only diuretic for which ≥1 drug-specific molecular target(s) were found to overlap in the PAH module but was prescribed too infrequently (n=1) in this study population for further analysis. Among participants prescribed diuretics exclusive of spironolactone at the time of RHC, there was no relationship identified between therapeutic target- interactome alignment and 6MWD, BNP levels, or WHO-FC at 6- or 12-months (data not shown). Similarly, there were no associations identified between diuretic target-interactome alignment and REVEAL 2.0 risk scores at 6- or 12-months (data not shown). The models for the COMPERA 2.0 four- strata risk assessment were not able to converge due to small numbers.

[0213] Among those prescribed serotonin psychotropic drugs, increasing alignment may have been associated with worsening REVEAL 2.0 scores. As alignment increased, REVEAL 2.0 Risk scores tended to worsen from low to high risk for mortality at 6 months (p = 0.034), a relationship which was attenuated at 12 months (p = 0.18). See FIG. 24, which shows the relationship between serotonin- psychotropic comparator-target alignment and United States Registry to Evaluate Early and Long-Term PAH Disease Management (REVEAL) 2.0 risk scores at six (FIG. 24A) and twelve (FIG. 24B) months after right heart catheterization.

[0214] Conversely, participants that were prescribed non-serotonin psychotropics at the time of RHC and had target alignment were significantly more likely to have lower REVEAL 2.0 Risk scores at 6 months, such that alignment in gene targets was associated with change from high risk (≥9) to intermediate risk (7-8) (p = 0.050), a relationship which was not observed at 12 months (p = 0.59). Gene target-non-serotonin psychotropic alignment had no detectable relationship with other PAH end points.

[0215] Example 9: Systems Pharmacology.

[0216] PAH drug target-interactome pairs are produced by first searching Drugbank for the targets of PAH drugs and controls, identifying drug target-interactome pairs to produce patient-specific networks for the patients, and validating the PAH target (PAH module) in order to quantitate the proportion of total drug targets (N) / N in an individualized interactome.Attorney Docket No. 15024-398PC0

[0217] Drug-interactome pairs predict PAH treatment response. Diuretics outside of the PAH module have no relationship to PAH end points. Specifically, clinical effects of diuretics with targets outside of the PAH model are not due to affecting molecular pathways that are implicated in PAH pathobiology. See FIG. 26.

[0218] Example 10: Development of Individualized Interactomes (HCM).

[0219] To contextualize the complete PPI network of an individual patient, a strategy was modeled after the sample-statistical network method by Liu and colleagues. Specifically, a control correlation matrix was established by calculating the Pearson correlation coefficient of each protein pair across all control samples. See FIG. 27A. The Pearson correlation coefficient (r) is established for all protein (p) pairs using proteomics data from healthy controls. n = total combination of pairwise correlations; m = total number of proteins.

[0220] Next, the proteomic profile from a single patient sample was added to the control matrix, and the Pearson correlation coefficients are recalculated for all protein pairs. See FIG. 27B. Each protein pair was analyzed to determine if the addition of a single patient proteomic profile to the control matrix resulted in a significantly different Pearson correlation coefficient. This calculation was repeated for each protein pair ad seriatim. Only protein pairs with a Pearson correlation coefficient that was significantly different after the addition of a single patient proteomic profile were analyzed further. See FIG. 27C, step 1.

[0221] After Bonferroni correction for multiple comparisons, retained protein pairs were mapped to the consolidated human interactome. See FIG. 27C, steps 2-3. If the protein pair was identified in the consolidated human interactome, this indicated a functional or physical interaction between the protein pair. Using this method, a PPI network was then established, which represents the patient-specific dysfunctional or perturbed biological system compared to data from controls. See FIG. 28C, step 4. See the examples below, where evidence in support of this approach for identifying patient-specific reticulotypes is provided.

[0222] The results for developing patient-specific protein-protein interaction networks in PAH are shown in FIG. 27. FIG 27A shows significant correlations in patients. FIG. 27B is the proteomic profile of an individual patient, which is then added to the control correlation matrix, and the new Pearson correlation coefficient (r') is calculated for all protein pairs. The patient proteomic profile is then removed from the matrix and the process is repeated sequentially for all patients. FIG. 27C, step 1,Attorney Docket No. 15024-398PC0 shows collection of statistically significant changes between the collected r and r' coefficients. FIG. 27C, step 2, shows those protein pairs (P1–P2), which were mapped to the human interactome (Step 3), which contains information on 15,489 proteins and 188,973 protein-protein interactions (PPIs). FIG. 27C, step 4, shows the protein pairs for which a PPI was identified in the human interactome, used to generate individual-patient PPI networks.

[0223] Example 11: Analyzing Temporal Changes in the Patient-specific PPI Networks.

[0224] Most studies linking big datasets to clinical outcome in PAH rely on single assay measurements. Analyzing changes in biological data over time is a more powerful strategy, but such data are generally not available. The paired samples available in the PAH Phenome Biobank allow one to postulate that monitoring stable and dynamic network features over time will glean novel and important insight on PPI interactions that characterize PAH pathobiology longitudinally. Studying patient-specific PPI networks is able to identify functionally important molecular pathways that emerge temporally, and associate them with clinical risk or resilience in individual patients.

[0225] Example 12: Patient-specific PPI Networks from Asymptomatic Relatives of PAH Patients.

[0226] We chose to first test the method in hypertrophic cardiomyopathy (HCM) based on four features of HCM that overlap with PAH: i) heterogenous pathobiology across patients, ii) pathology that is defined by a diverse range of endophenotypes, iii) evidence suggesting that reductionist analytical approaches may overlook important, but integrated pathogenetic mechanisms and iv) increased mPAP and PVR.

[0227] Asymptomatic relatives of patients with a PAH-associated genetic mutation are at increased risk of developing PAH themselves. The role of genetic context in predicting PAH incidence is supported by covering lines of evidence from basic, clinical and population studies. However, most ‘PAH- causing’ genotypes are incompletely penetrant, raising speculation that genetic co-regulation or integrated signaling pathways may be important in understanding HPAH pathogenesis. To address this knowledge gap, samples from asymptomatic relatives of PAH patients as well as healthy controls can be studied. The methodology used here can decipher the molecular basis of HPAH by (i) characterizing the presence (or absence) of putative PAH genes in the patient-specific PPI networks and (ii) determining if common PPI pathways emerge among patients sharing the same genotype. This individualizedAttorney Docket No. 15024-398PC0 information on pathways that underpin HPAH in affected patients and at-risk family members can help to identify persons more likely to develop PAH.

[0228] Example 13: Patient-specific PPI networks.

[0229] Patient-specific PPI networks were used to clarify complex phenotypes to advance precision medicine. Left ventricular (LV) tissue from health controls (rejected heart transplant donors) (N=5) and patients with HCM (N=18) were used for these analyses (IRB #9487). HCM is a disease of increased LV thickening and numerous other cardiac structural abnormalities not due to a secondary cause (e.g., hypertension). Mutations in genes coding for cardiomyocyte sarcomere proteins predict familial disease inheritance in some patients and account for selected HCM pathobiological mechanisms, but do not explain the totality of HCM phenotypic features (similar to reports for genetic variants in PAH).

[0230] FIG. 28 shows the variance in the transcriptomic profile of patients with hypertrophic cardiomyopathy compared to controls. To obtain a transcriptomic profile, RNA-Seq was performed on mRNA isolated from control and HCM LV tissue samples. The count matrix included 44,287 genomic features from 23 samples; the transcript alignment by Tophat2 was 90.3 ± 1.0 % for the total cohort, with no significant differences observed between HCM and controls (90.0 vs. 91.7 %, P=0.49). We identified 1,658 significantly differentially expressed transcripts between the control and HCM cohorts; the expression pattern for individual patients is presented in heat map form. See results in FIG. 28A. For FIG. 28A, left ventricular (LV) myocardial samples from healthy controls (N=5) and myectomy samples from hypertrophic cardiomyopathy (HCM) patients (N=18) were analyzed by RNA-Seq. Differentially expressed transcripts between groups are presented as a heat-map (N=2,238).

[0231] FIG. 28B shows the variance in transcriptomic profile for controls and HCM patients for the 1st and 2nd principal component (PC) vectors. Partitioning variance in gene expression by the first and second principal component vectors accounted for only 47% of the total variance, but demonstrated separation between patient groups as well as substantial separation among patients within the HCM group (FIG. 28B). These data suggest substantial differences in the transcriptomic profile between controls vs. HCM, as well as across individual patients HCM and imply that analytical methods using averaged value are likely to overlook potentially important pathobiological information in individual patients, emphasizing the benefits of patient-specific network medicine.

[0232] To investigate whether individual-patient networks can distinguish HCM from other forms of cardiomyopathy, we performed tests to determine if our network analysis could discriminate theAttorney Docket No. 15024-398PC0 transcriptomic profile of patients with HCM from dilated cardiomyopathy (DCM). This analysis was pursued because although HCM and DCM are distinct phenotypically, there are accumulating data suggesting that numerous identical ‘causative’ genetic variants underpin both of these two diseases. From the RNA-Seq data, 26% of all differentially expressed genes between DCM vs. control were also observed in HCM patients vs. control. Thus, PPIs apparently were able to delineate HCM from DCM. To test this, individual DCM patient networks were generated as was done for HCM. Although node overlap was substantial between HCM and DCM (0.55 ± 0.03), there was virtually no overlap among PPIs (e.g., network edges) between these two diseases (0.03 ± 0.0) (FIG. 29B).

[0233] Using methods outlined above, patient-specific PPI networks were formed (n = 18). These networks varied widely in topology (3,629 (2,474 min, 5,031 max) number of nodes) and complexity (3,696 (2014 min, 6,626 max) number of edges) (see FIG. 29A). To determine overlap between networks for edges or nodes, the modified Jaccard index was used. A significantly greater overlap for nodes was observed compared to edges across the HCM networks (0.73 ± 0.04 vs. 0.48 ± 0.09, P=3.2e- 82). See FIG. 29B.

[0234] These data suggest that PPIs are important for distinguishing the pathobiology individual patients in silico. The PPIs fully delineated HCM from DCM. Overall, the data in this example demonstrate that focusing on network connectivity is a powerful strategy by which to delineate the pathobiology of different cardiovascular diseases as well as patients with the same complex disease.

[0235] Example 14: Leveraging Genetic Context.

[0236] In order to leverage genetic context in patient-specific PPI networks to identify disease endophenotypes, we next tested the postulate that sarcomere protein-coding genes harboring reported HCM pathogenic variants would be central to PPIs in individual HCM patient networks. Among 17 reported ‘HCM-causing’ genes, including 774 pathogenic and 475 likely pathogenic variants, there were 4 variants in 2 HCM genes affecting a total of 4 patients (N=25%): 3 patients had a pathogenic mutation (MYBPC3: c.1928-2A>G; MYBPC3: c.G2497A:p.A833T; MYL2: c.C141A:p.N47K) and one patient had a likely pathogenic mutation (MYL: c.C141A:p.N47K). However, nodes corresponding to these genes were not present in any of the patient-specific PPI networks from these four patients. This indicated that the methods discussed here can be used to discover alternative molecular interactions important in HCM. Genes associated with 54 representative endophenotypes that might be relevant to HCM were uncovered.Attorney Docket No. 15024-398PC0

[0237] The hypergeometric test was used next to determine if a patient-specific PPI network was significantly enriched with any of these 54 endophenotypes compared to the human interactome as the background. This analysis identified 32 endophenotypes that were significantly enriched in each patient networks across the entire HCM cohort, among which N=25 had not been reported previously in this disease. The heatmap in FIG. 30 shows an endophenotype enrichment profile for patient-specific PPI networks. In the heatmap, the rows correspond to the endophenotype classifications, and the columns denote each HCM patient. This approach identified N=25 novel HCM endophenotypes.

[0238] Example 15: Patient-specific PPI Networks for Developing Individualized Reticulotypes.

[0239] Focusing on fibrosis, we observed that a greater proportion of patient-specific networks were enriched for fibrosis compared with hypertrophy pathways (94% vs. 11%, 1.3e-05, N=18). See FIG. 31. In addition, increased interstitial collagen was observed by trichrome staining in the HCM myectomy samples (4.7 ± 0.42 vs. 14.3 ± 0.02 % collagen / high powered field, P=0.04, N=3-13 / condition), but no significant difference in cardiomyocyte cross-sectional diameter was observed compared with controls (P=0.57). This finding is an example of how patient-specific PPI networks can inform major conceptual paradigm shifts. In this case, demonstrating that LV thickening in HCM may be due to matrix remodeling rather than hypertrophic signaling. Compared to hypertrophy pathways, the proportion of HCM patients with networks enriched for fibrosis signaling pathways was significantly greater. See FIG. 31 (HCM, hypertrophic cardiomyopathy).

[0240] Whether patient-specific network features were useful for phenotyping patients clinically was evaluated. The number of unique fibrosis edges in each patient-specific PPI network was useful for predicting extreme cardiac fibrotic phenotypes, see FIG. 32, and correlated with prognostic cardiac structural and hemodynamic variables including: LV end-diastolic volume (r= - 0.67, P=0.025), cardiac output (r= -0.60, P<0.05), and indexed PVR (r= 0.76, P=0.010). FIG. 32 shows the fibrosis burden in silico and fibrosis phenotype. FIG. 32A: the number of fibrosis nodes in edges unique to a single HCM patient network, provided in descending order, suggests that HCM patient 8 and HCM patient 7 had the greatest and least network fibrosis burden, respectively. This was confirmed by trichrome and picrosirius stain performed on LV cardiac tissues. FIG. 32B: a positive correlation was observed between the number of unique fibrosis edges for each patient network and interstitial collagen quantity by trichrome stain across all HCM patients, which was maintained even after excluding the most densely connected networks for fibrosis (N=3 patients) (inset). Representative micrographs and blots are shown.Attorney Docket No. 15024-398PC0 REFERENCES

[0241] All publications listed below and throughout the specification are hereby incorporated by reference in their entirety. Nothing herein is to be construed as an admission that the invention is not entitled to antedate such disclosure by virtue of prior invention. 1. Aghamohammadzadeh R… Maron BA. Upregulation of the Mammalian Target of Rapamycin Complex 1 Subunit Raptor by Aldosterone Induces Abnormal Pulmonary Artery Smooth Muscle Cell Survival Patterns to Promote Pulmonary Arterial Hypertension. FASEB J. 2016;30(7):2511-2517. PMCID: PMC4904292. 2. Apweiler, R. et al. UniProt: the Universal Protein knowledgebase. Nucleic Acids Res. 32, D115– D119 (2004). 3. Aranda B, et al. The IntAct molecular interaction database in 2010. Nucleic Acids Res. 2010;38:D525-31. 4. Barabási AL, Gulbahce N, Loscalzo J. Network medicine: a network-based approach to human disease. Nat Rev Genet 2011;12:56-68. PMCID: PMC3140052. 5. Benza RL, et al. Predicting survival in pulmonary arterial hypertension: insights from the Registry to Evaluate Early and Long-Term Pulmonary Arterial Hypertension Disease Management (REVEAL). Circulation 2010; 122: 164–72. 6. Benza RL, Gomberg-Maitland M, Elliott CG, et al. Predicting survival in patients with pulmonary arterial hypertension: The REVEAL Risk Score Calculator 2.0 and comparison with ESC / ERS-based risk assessment strategies. Chest 2019; 156: 323-337. 2019 / 02 / 18. DOI: 10.1016 / j.chest.2019.02.004. 7. Boucly A, Weatherald J, Savale L, et al. External validation of a refined four-stratum risk assessment score from the French pulmonary hypertension registry. Euro Respir J 2022; 5960: 2102419. DOI: 10.1183 / 13993003.02419-2021. 8. Budhiraja R, Tuder RM and Hassoun PM. Endothelial dysfunction in pulmonary hypertension. Circulation 2004; 109: 159-165. 2004 / 01 / 22. DOI: 10.1161 / 01.cir.0000102381.57477.50. Cahill E, et al. Gremlin plays a key role in the pathogenesis of pulmonary hypertension. Circulation 2012;125(7):920-30. PMID: 22247494. 9. Chelladurai P, Savai R and Pullamsetti SS. Zooming into cellular and molecular heterogeneity of pulmonary hypertension. what more single-cell omics can offer. Am J Respir Crit Care Med 2021; 203: 941-943. DOI: 10.1164 / rccm.202010-3889ED.Attorney Docket No. 15024-398PC0 10. Ceol A, et al. MINT, the molecular interaction database: 2009 update. Nucleic Acids Res 2010;38:D532-9. 11. Cheng F…Loscalzo J. Network-based approach to prediction and population-based validation of in silico drug repurposing. Nat Commun 2018;9:2691. PMCID: PMC6043492. 12. Cheng F…Loscalzo J. A genome-wide positioning systems network algorithm for in silico drug repurposing. Nat Commun 2019;10:3476. PMCID: PMC6677722. 13. Courboulin A et al. Role for miR-204 in human pulmonary arterial hypertension. J Exp Med. 2011 Mar 14;208(3):535-48. 14. Covella M…Maron BA. Mechanism of progressive heart failure and significance of pulmonary hypertension in obstructive hypertrophic cardiomyopathy. Circ Heart Fail 10, e003689 (2017). PMCID: PMC5390526. 15. Dardi F, et al. Combining bosentan and sildenafil in pulmonary arterial hypertension patients failing monotherapy: real-world insights. Eur Respir J. 2015 Aug;46(2):414-21. 16. Delcroix M, Howard L. Pulmonary arterial hypertension: the burden of disease and impact on quality of life. Eur Respri Rev 2015;24:621-29. 17. Dunning LT, Lundgren MR, Moreno-Villena JJ, et al. Introgression and repeated co-option facilitated the recurrent emergence of C(4) photosynthesis among close relatives. Evolution 2017; 71: 1541-1555. 20170428. DOI: 10.1111 / evo.13250. 18. Dweik RA, Rounds S, Erzurum SC, et al. An Official American Thoracic Society Statement: Pulmonary Hypertension Phenotypes. Am J Respir Crit Care Med 2014; 189: 345-355. DOI: 10.1164 / rccm.201311-1954ST. 19. European Pulmonary Hypertension Association (PHA). The impact of pulmonary arterial hypertension (PAH) on the lives of patients and carers: results from an international survey. www.phaeurope.org / wp- content / uploads / International-PAH-patient-and-Carer-Survey-Report- FINAL1.pdf Date last updated: September 2012. Date last accessed: June 29, 2015. 20. Evans CE, Cober ND, Dai Z, et al. Endothelial cells in the pathogenesis of pulmonary arterial hypertension. Eur Respir J 2021; 5820210902. DOI: 10.1183 / 13993003.03957-2020. 21. Evans JD, et al. BMPR2 mutations and survival in pulmonary arterial hypertension: an individual participant data meta-analysis. Lancet Respir Med. 2016 Feb;4(2):129-37. 22. Freeman LC. A set of measures of centrality based on betweenness. Sociometry. 1977;40:35-41. 23. Frid MG, et al. Immunoglobulin-Driven Complement Activation Regulates Pro-InflammatoryAttorney Docket No. 15024-398PC0 Remodeling in Pulmonary Hypertension. Am J Respir Crit Care Med. 2019 Sep 23. doi: 10.1164 / rccm.201903-0591OC. [Epub ahead of print]. 24. Galiè N, et al. Initial Use of Ambrisentan plus Tadalafil in Pulmonary Arterial Hypertension. N Engl J Med. 2015;373(9):834–844. 25. Gaulton, A. et al. ChEMBL: a large-scale bioactivity database for drug discovery. Nucleic Acids Res. 40, D1100–D1107 (2012). 26. Ghofrani HA, et al. Riociguat for the treatment of pulmonary arterial hypertension. N Engl J Med. 2013;369(4):330–340. 27. Graf S, et al. Identification of rare sequence variation underlying heritable pulmonary arterial hypertension. Nat Commun 2018;9:1416. 28. Harbaum L, Rhodes CJ, Wharton J, et al. Mining the plasma proteome for insights into the molecular pathology of pulmonary arterial hypertension. Am J Respir Crit Care Med 2022; 205: 1449- 1460. DOI: 10.1164 / rccm.202109-2106OC. 29. Harzheim D, Klose H, Pinado FP, et al. Anxiety and depression disorders in patients with pulmonary arterial hypertension and chronic thromboembolic pulmonary hypertension. Respir Res 2013; 14: 104. 20131009. DOI: 10.1186 / 1465-9921-14-104. 30. Havugimana PC, et al. A census of human soluble protein complexes. Cell. 2012;50:1068-81. Morrell NW, et al. Genetics and genomics of pulmonary arterial hypertension. Eur Respir J. 2019 Jan 24;53(1). pii: 1801899. PMCID: PMC6351337. 31. Hemnes AR, et al. PVDOMICS: A Multi-Center Study to Improve Understanding of Pulmonary Vascular Disease Through Phenomics. Circ Res. 2017 Oct 27;121(10):1136-1139. PMCID: PMC5685561. 32. Hernandez-Boussard, T. et al. The pharmacogenetics and pharmacogenomics knowledge base: accentuating the knowledge. Nucleic Acids Res. 36, D913–D918 (2008). 33. Hinson JT, et al. Science 349:982-6, 2015. 34. Hoeper MM, Dwivedi K, Pausch C, et al. Phenotyping of idiopathic pulmonary arterial hypertension: a registry analysis. Lancet Respir Med 2022; 10: 937-948. 20220628. DOI: 10.1016 / s2213-2600(22)00097-2. 35. Hoeper MM, Pausch C, Grünig E, et al. Idiopathic pulmonary arterial hypertension phenotypes determined by cluster analysis from the COMPERA registry. J Heart Lung Transplant 2020; 39: 1435- 1444. 20200930. DOI: 10.1016 / j.healun.2020.09.011.Attorney Docket No. 15024-398PC0 36. Hoeper MM, Badesch DB, Ghofrani HA, et al. Phase 3 trial of sotatercept for treatment of pulmonary arterial hypertension. New Engl J Med 2023; 388: 1478-1490. DOI: 10.1056 / NEJMoa2213558. 37. Hornbeck PV, et al. PhosphoSitePlus: a comprehensive resource for investigating the structure and function of experimentally determined post-translational modifications in man and mouse. Nucleic Acids Res. 2012; 40:D261-70. 38. Hossie TJ, Gobin J and Murray DL. Confronting missing ecological data in the age of pandemic lockdown. Front Ecol Evol 2021. 39. Hsu S, et al. Right ventricular functional reserve in pulmonary arterial hypertension. Circulation 2016;133(24):2413-22. 40. Humbert M, McLaughlin V, Gibbs JSR, et al. Sotatercept for the treatment of pulmonary arterial hypertension. New Eng J Med 2021; 384: 1204-1215. DOI: 10.1056 / NEJMoa2024277. 41. Humbert M, Kovacs G, Hoeper MM, et al. 2022 ESC / ERS Guidelines for the diagnosis and treatment of pulmonary hypertension: Developed by the task force for the diagnosis and treatment of pulmonary hypertension of the European Society of Cardiology (ESC) and the European Respiratory Society (ERS). Endorsed by the International Society for Heart and Lung Transplantation (ISHLT) and the European Reference Network on rare respiratory diseases (ERN-LUNG). Eur Heart J 2022; 43: 3618-3731. DOI: 10.1093 / eurheartj / ehac237. 42. Humbert M, et al. Survival in patients with idiopathic, familial, and anorexigen-associated pulmonary arterial hypertension in the modern management era. Circulation 2010;122:156-63. PMID: 20585011. 43. Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M. The KEGG resource for deciphering the genome. Nucleic Acids Res, 2004;32:D277-80. 44. Kariotis S, Jammeh E, Swietlik EM, et al. Biological heterogeneity in idiopathic pulmonary arterial hypertension identified through unsupervised transcriptomic profiling of whole blood. Nat Commun 2021; 12: 7104. 20211207. DOI: 10.1038 / s41467-021-27326-0. 45. Keshava et al. Human Protein Reference Database--2009 update. Nucleic Acids Res. 2009;37:D767-72. 46. Kim D, Langmead B and Salzberg SL. HISAT: a fast spliced aligner with low memory requirements. Nature Methods 2015; 12: 357-360. DOI: 10.1038 / nmeth.3317.Attorney Docket No. 15024-398PC0 47. Koscielny G, An P, Carvalho-Silva D, et al. Open Targets: a platform for therapeutic target identification and validation. Nucleic Acids Res 2017; 45: D985-d994. 20161129. DOI: 10.1093 / nar / gkw1055. 48. Law, V. et al. DrugBank 4.0: shedding new light on drug metabolism. Nucleic Acids Res. 42, D1091–D1097 (2014). 49. Lazarus HM, Denning J, Wring S, et al. A trial design to maximize knowledge of the effects of rodatristat ethyl in the treatment of pulmonary arterial hypertension (ELEVATE 2). Pulm Circ 2022; 12: e12088. 20220511. DOI: 10.1002 / pul2.12088. 50. Lee LY, Loscalzo J. Network medicine in pathobiology. Am J Pathol 2019;189, 1311-1326. PMCID: PMC6616999. 51. Lee DS, Park J, Kay KA, Chirstakis NA, Oltvai ZN, Barabasi AL. The implications of huma metabolic network topology for disease comorbidity. Proc Natl Acad Sci U S A. 2008; 105: 9880-5. 52. Leopold JA. PVDOMICS proteomic profile. 2019 American College of Cardiology Scientific Sessions, New Orleans, LA (Abstract). 53. Leopold J, Maron BA, and Loscalzo J. The Application of big data to cardiovascular disease: paths to precision medicine. Journal of Clinical Investigation. 2020 Jan 2;130(1):29-38. PMCID: PMC6934200. 54. Lipke DW, et al. Temporal alterations in specific basement membrane components in lungs from monocrotaline-treated rats. Am J Respir Cell Mol Biol. 1993 Oct;9(4):418-28. 55. Liu X, Wang Y, Ji H, et al. Personalized characterization of diseases using sample-specific networks. Nucleic Acids Res 2016; 44: e164. 20160904. DOI: 10.1093 / nar / gkw772. 56. Liu, T. Q., Lin, Y. M., Wen, X., Jorissen, R. N. & Gilson, M. K. BindingDB: a web-accessible database of experimentally determined protein-ligand binding affinities. Nucleic Acids Res. 35, D198– D201 (2007). 57. Mainguy V, et al. Peripheral muscle dysfunction in idiopathic pulmonary arterial hypertension. Thorax. 2010 Feb;65(2):113-7. 58. Manolio TA, et al. Finding the missing heritability of complex diseases. Nature 2009;461(7265):747-53. 59. NCT01185730 on clinicaltrials.gov 60. Maron B.J., et al. JAMA 301:1253-9, 2009. 61. Maron, B.J., Maron, M.S. & Semsarian, C. Genetics of hypertrophic cardiomyopathy after 20 years:Attorney Docket No. 15024-398PC0 clinical perspectives. J Am Coll Cardiol 60, 705-15 (2012). PMID: 22796258. 62. Maron BJ, Maron MS, Maron BA, Loscalzo, J. Moving beyond the sarcomere to explain heterogeneity in hypertrophic cardiomyopathy. J Am Coll Cardiol 73, 1978-86 (2019). PMCID: PMC6550351. 63. Maron BA, Galie N. Diagnosis, Treatment, and Clinical Management of Pulmonary Arterial Hypertension in the Contemporary Era: A Review. JAMA Cardiol. 2016 Dec 1;1(9):1056-1065. PMCID: PMC5177491. 64. Maron BA, et al. Aldosterone increases oxidant stress to impair guanylyl cyclase activity by cysteinyl thiol oxidation in vascular smooth muscle cells. J Biol Chem. 2009;284(12):7665-72. PMCID: PMC2658060. 65. Maron BA, et al. Aldosterone inactivates the endothelin-B receptor to decrease pulmonary endothelial nitric oxide levels and modulate pulmonary arterial hypertension. Circulation. 2012;126(8):963-974. PMCID: PMC3534848. 66. Maron BA, et al. Upregulation of steroidogenic acute regulatory protein by hypoxia stimulates aldosterone synthesis in pulmonary artery endothelial cells to promote pulmonary vascular fibrosis. Circulation. 2014;130(2):168-172. PMCID: PMC4090713. 67. Maron BA, Wang RS, Shevtsov S, et al. Individualized interactomes for network-based precision medicine in hypertrophic cardiomyopathy with implications for other clinical pathophenotypes. Nat Commun 2021; 12: 873. 20210208. DOI: 10.1038 / s41467-021-21146-y. 68. Maron BA, et al. Plasma aldosterone levels are elevated in patients with pulmonary arterial hypertension in the absence of left ventricular heart failure: a pilot study. Eur J Heart Fail. 2013 Mar;15(3):277-83. PMCID: PMC3576899. 69. Maron BA. Inhibiting serotonin to treat pulmonary arterial hypertension. JACC: Basic Transl Sci 2023; 8: 1389-1391. DOI: doi:10.1016 / j.jacbts.2023.09.008. 70. McCollister DH, Beutz M, McLaughlin V, et al. Depressive symptoms in pulmonary arterial hypertension: prevalence and association with functional status. Psychosomatics 2010; 51: 339- 339.e338. DOI: 10.1176 / appi.psy.51.4.339. 71. McNally EM, et al. The genetic landscape of cardiomyopathy and its role in heart failure. 2015;Cell Metab 21, 174-82. 72. Menche J, et al. Disease networks. Uncovering disease-disease relationships through the incomplete interactome. Science 2015;347(6224):1247601. PMCID: PMC4435741.Attorney Docket No. 15024-398PC0 73. Min J, Badesch D, Chakinala M, et al. Prediction of health-related quality of life and hospitalization in pulmonary arterial hypertension: The Pulmonary Hypertension Association Registry. Am J Respir Crit Care Med 2021; 203: 761-764. DOI: 10.1164 / rccm.202010-3967LE. 74. Montani D, et al. Am J Respir Crit Care Med 198:537-541, 2018. 75. Morrell NW. Role of bone morphogenetic protein receptors in the development of pulmonary arterial hypertension. Adv Exp Med Biol. 2010;661:251-64. PMID:20204735. 76. Morrell NW, et al. Genetics and genomics of pulmonary arterial hypertension. Eur Respir J. 2019 Jan 24;53(1). PMCID: PMC6391516. 77. Moutchia J, McClelland RL, Al-Naamani N, et al. Pulmonary arterial hypertension treatment: an individual participant data network meta-analysis. Eur Heart J 2024; 45: 1937-1952. DOI: 10.1093 / eurheartj / ehae049. 78. Moutchia J, McClelland RL, Al-Naamani N, et al. Minimal clinically important difference in the 6- minute-walk distance for patients with pulmonary arterial hypertension. Am J Respir Crit Care Med 2023; 207: 1070-1079. DOI: 10.1164 / rccm.202208-1547OC. 79. Naeije R and Eddahibi S. Serotonin in pulmonary arterial hypertension. Am J Respir Crit Care Med 2004; 170: 209-210. DOI: 10.1164 / rccm.2406002. 80. Oldham WM…Maron BA. Network Analysis to Risk Stratify Patients With Exercise Intolerance. Circ Res. 2018 Mar 16;122(6):864-876. PMCID: PMC5924425. 81. Ortega, A, et al. Patients with Dilated Cardiomyopathy and Sustained Monomorphic Ventricular Tachycardia Show Up-Regulation of KCNN3 and KCNJ2 Genes and CACNG8-Linked Left Ventricular Dysfunction. PLoS One, 2015;10: e0145518 (2015). 82. Pan HM, McClelland RL, Moutchia J, et al. Heterogeneity of treatment effects by risk in pulmonary arterial hypertension. Eur Respir J 2023; 6220230727. DOI: 10.1183 / 13993003.00190-2023. 83. Parikh VN, et al. MicroRNA-21 integrates pathogenic signaling to control pulmonary hypertension: results of a network bioinformatics approach. Circulation. 2012 Mar 27;125(12):1520-32. PMCID: PMC3353408 / 84. Pawson, A. J. et al. The IUPHAR / BPS Guide to PHARMACOLOGY: an expert-driven knowledgebase of drug targets and their ligands. Nucleic Acids Res. 42, D1098–D1106 (2014). 85. Piñero J, Bravo À, Queralt-Rosinach N, et al. DisGeNET: a comprehensive platform integrating information on human disease-associated genes and variants. Nucleic Acids Res 2017; 45: D833-d839. 20161019. DOI: 10.1093 / nar / gkw943.Attorney Docket No. 15024-398PC0 86. Pollett JB, Benza RL, Murali S, et al. Harvest of pulmonary artery endothelial cells from patients undergoing right heart catheterization. J Heart Lung Transplant 2013; 32: 746-749. 2013 / 05 / 21. DOI: 10.1016 / j.healun.2013.04.013. 87. Pulido T, et al. Macitentan and morbidity and mortality in pulmonary arterial hypertension. N Engl J Med. 2013;369(9):809–818. 88. Rhodes CJ, et al. Genetic determinants of risk in pulmonary arterial hypertension: international genome- wide association studies and meta-analysis. Lancet Respir Med.2019 Mar;7(3):227-238. 89. Rhodes CJ, et al. Plasma proteome analysis in patients with pulmonary arterial hypertension: an observational study. Lancet Respir Med 2017;5:717-26. PMCID: PMC5573768. 90. Rhodes CJ, et al. Plasma metabolomics implicates modified transfer RNAs and altered bioenergetics in thee outcomes of pulmonary arterial hypertension. Circulation 2017;135:460-75. PMCID: PMC5287439. 91. Rhodes CJ, Ghataorhe P, Wharton J, et al. Plasma metabolomics implicates modified transfer rnas and altered bioenergetics in the outcomes of pulmonary arterial hypertension. Circulation 2017; 135: 460-475. 2016 / 11 / 25. DOI: 10.1161 / circulationaha.116.024602. 92. Rhodes CJ, Batai K, Bleda M, et al. Genetic determinants of risk in pulmonary arterial hypertension: international genome-wide association studies and meta-analysis. Lancet Respir Med 2019; 7: 227-238. 20181205. DOI: 10.1016 / s2213-2600(18)30409-0. 93. Rhodes CJ, Otero-Núñez P, Wharton J, et al. Whole-blood RNA profiles associated with pulmonary arterial hypertension and clinical outcome. Am J Respir Crit Care Med 2020; 202: 586-594. DOI: 10.1164 / rccm.202003-0510OC. 94. Rich S, et al. The effect of high doses of calcium-channel blockers on survival in primary pulmonary hypertension. N Engl J Med. 1992 Jul 9;327(2):76-81. 95. Ruepp A, et al. CORUM: the comprehensive resource of mammalian protein complexes--2009. Nucleic Acids Res. 2010;38: D497-501. 96. Ryanto GRT, Ikeda K, Miyagawa K, et al. An endothelial activin A-bone morphogenetic protein receptor type 2 link is overdriven in pulmonary hypertension. Nat Commun 2021; 12: 1720. DOI: 10.1038 / s41467-021-21961-3. 97. Sedaghat F, et al. EHJ 38:3449-60, 2017. 98. Sadoughi A, Roberts KE, Preston IR, et al. Use of selective serotonin reuptake inhibitors and outcomes in pulmonary arterial hypertension. Chest 2013; 144: 531-541. DOI: 10.1378 / chest.12-2081.Attorney Docket No. 15024-398PC0 99. Samokhin AO…Maron BA. Circulating NEDD9 is increased in pulmonary arterial hypertension: A multicenter, retrospective analysis. J Heart Lung Transpl; 2020, In press. PMID: 31952977. 100. Samokhin AO, et al. Isolating pulmonary microvascular endothelial cells ex vivo: Implications for pulmonary arterial hypertension, and a caution on the use of commercial biomaterials. Sci Transl Med 2018;10(445). Pii: eaap7294. PMCID: PMC6392245. 101. Sikirica M, et al. The economic burden of pulmonary arterial hypertension (PAH) in the US on payers and patients. BMC Health Serv Res. 2014 Dec 24;14:676. 102. Simonneau G, Gatzoulis MA, Adatia I, et al. Updated clinical classification of pulmonary hypertension. J Am Coll Cardiol 2013; 62: 029. Research Support, NIH , Extramural Research Support, Non-U S Gov't Review. 103. Simonneau G, et al. Haemodynamic definitions and updated clinical classification of pulmonary hypertension. Eur Respir J. 2019 Jan 24;53(1). pii: 1801913. 104. Singh N, Eickhoff C, Garcia-Agundez A, et al. Transcriptional profiles of pulmonary artery endothelial cells in pulmonary hypertension. Sci Rep 2023; 13: 22534. 20231218. DOI: 10.1038 / s41598- 023-48077-6. 105. Sitbon O, et al. Selexipag for the treatment of pulmonary arterial hypertension. N Engl J Med. 2015;373(26):2522–2533. 106. Soon E, Holmes AM, Treacy CM, et al. Elevated levels of inflammatory cytokines predict survival in idiopathic and familial pulmonary arterial hypertension. Circulation 2010; 122: 920-927. 2010 / 08 / 18. DOI: 10.1161 / circulationaha.109.933762. 107. Soubrier F, et al. Genetics and genomics of pulmonary arterial hypertension. J Am Coll Cardiol. 2013 Dec 24;62(25 Suppl):D13-21. 108. Stark C, et al. The BioGRID Interaction Database: 2011 update. Nucleic Acids Res. 2011; 39:D698-704. 109. Stitziel NO, et al. NEJM 371:2072-82, 2014. 110. Studer S, et al. Treatment patterns, healthcare resource utilization, and healthcare costs among patients with pulmonary arterial hypertension in a real-world US database. Pulm Circ. 2019 Jan- Mar;9(1):2045894018816294. 111. Sweatt AJ, Hedlin HK, Balasubramanian V, et al. Discovery of distinct immune phenotypes using machine learning in pulmonary arterial hypertension. Circ Res 2019; 124: 904-919. 2019 / 01 / 22. DOI: 10.1161 / circresaha.118.313911.Attorney Docket No. 15024-398PC0 112. Taichman DB, et al. Health-related quality of life in patients with pulmonary arterial hypertension. Respir Res 2005; 6:92. 113. Thenappan T, Pulmonary arterial hypertension: pathogenesis and clinical management. Br Med J 2018 Mar 14;360:j5492. 114. Tian W, et al. Phenotypically Silent Bone Morphogenetic Protein Receptor 2 Mutations Predispose Rats to Inflammation-Induced Pulmonary Arterial Hypertension by Enhancing the Risk for Neointimal Transformation. Circulation. 2019 Oct 22;140(17):1409-1425. 115. Tielemans B, Stoian L, Wagenaar A, et al. Incremental Experience in In Vitro Primary Culture of Human Pulmonary Arterial Endothelial Cells Harvested from Swan-Ganz Pulmonary Arterial Catheters. Cells 2021; 1020211119. DOI: 10.3390 / cells10113229. 116. Ventetuolo CE, Aliotta JM, Braza J, et al. Culture of pulmonary arterial endothelial cells from pulmonary artery catheter balloon tips: Considerations for use in pulmonary vascular disease. Eur Respir J 20202020 / 01 / 18. DOI: 10.1183 / 13993003.01313-2019. 117. Vidal M, Cusick ME and Barabási AL. Interactome networks and human disease. Cell 2011; 144: 986-998. DOI: 10.1016 / j.cell.2011.02.016. 118. Vijaymeena MK, et al A Survey on Similarity Measures in Text Mining: Machine Learning and Applications: An Int J 2016;3:19–28. 119. Vittinghoff E, et al. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710-718. 120. Walsh TP, Baird GL, Atalay MK, et al. Experimental design of the Effects of Dehydroepiandrosterone in Pulmonary Hypertension (EDIPHY) trial. Pulm Circ 2021; 11: 2045894021989554. 2021 / 06 / 08. DOI: 10.1177 / 2045894021989554. 121. Wang RS, Oldham WM and Loscalzo J. Network-based association of hypoxia-responsive genes with cardiovascular diseases. New J Phys 2014; 16: 105014. DOI: 10.1088 / 1367-2630 / 16 / 10 / 105014. 122. Wang RS and Loscalzo J. Network module-based drug repositioning for pulmonary arterial hypertension. CPT Pharmacometrics Syst Pharmacol 2021; 10: 994-1005. 20210709. DOI: 10.1002 / psp4.12670. 123. Wellbrock J, et al. Intrinsic BMP Antagonist Gremlin-1 as a Novel Circulating Marker in Pulmonary Arterial Hypertension. Lung. 2015 Aug;193(4):567-70. 124. Williams SA, et al. Plasma protein patterns as comprehensive indicators of health. Nat Med Nat Med. 2019 Dec;25(12):1851-1857.Attorney Docket No. 15024-398PC0 125. Yu W, Clyne M, Khoury MJ, et al. Phenopedia and Genopedia: disease-centered and gene- centered views of the evolving knowledge of human genetic associations. Bioinformatics 2010; 26: 145- 146. 20091027. DOI: 10.1093 / bioinformatics / btp618. 126. Zhu, F. et al. Therapeutic target database update 2012: a resource for facilitating target-oriented drug discovery. Nucleic Acids Res. 40,D1128–D1136 (2012). Zile MR, et al. Circulation 131:1247-59, 2012.

Claims

Attorney Docket No. 15024-398PC0 CLAIMS 1. A method of monitoring a complex disease defined by molecular heterogeneity across patients sharing the same diagnosis in a patient in need thereof, comprising: (a) obtaining a cell biopsy sample from tissue affected by the disease in the subject, wherein the sample is cells or a cell surrogate; (b) expanding cells in the cell biopsy sample in tissue culture; (c) isolating patient mRNA from the expanded cells and subjecting the mRNA to bulk RNA- SEQ; (d) compiling the transcript quantity for each gene in the patient mRNA to form a patient transcript profile and holding the profile for later analysis; (e) isolating control mRNA from cells of a normal control and subjecting the mRNA to bulk RNA-SEQ; (f) assembling the transcript quantity for each gene in each control subject in a matrix and calculating the Pearson correlation coefficient for each gene pair in the matrix; (g) adding the patient transcript profile to the matrix of (f); (h) recalculating the Pearson correlation coefficient for each gene pair in the matrix; (i) gathering the gene pairs with Pearson correlation coefficients that are statistically significantly different after the addition of the patient transcript profile to the matrix; (j) mapping the gene pairs of (i) to a consolidated human protein-protein interactome (PPI); (k) assembling the gene pairs that map to protein-protein interactions in the consolidated human PPI to produce a patient-specific PPI; and (l) monitoring the disease-specific gene expression and the clinical changes of the disease in the subject. 2 A method of claim 1 wherein the cell surrogate is selected from the group consisting of peripheral blood, serum, urine, cerebrospinal fluid, and saliva. 3 A method of claim 1 wherein the complex disease is a pulmonary vascular disease. 4 A method of claim 3 wherein the pulmonary vascular disease is pulmonary arterial hypertension.Attorney Docket No. 15024-398PC0 5. A method of claim 4 wherein the cell biopsy sample is a sample of pulmonary artery endothelial cells.

6. A method of claim 5 wherein the cell biopsy sample is obtained by right heart catheterization (RHC) and harvesting from the tips of RHC balloons. 7 A method of claim 1 wherein the complex disease is selected from the group consisting of cardiomyopathy; parenchymal lung disease; solid tumors affecting the breast, lung, colon, pancreas, or ovaries; inflammatory bowel disease; systemic lupus erythematosus and systemic sclerosis; mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, and metabolic dysfunction; and the like. 8 A method of claim 1 wherein the complex disease is Crohn’s disease or ulcerative colitis. 9 The method of claim 1 wherein the disease-specific gene expression signature and clinical changes in the subject are monitored to provide real-time information of the disease pathobiology. 10 A method of treating a complex disease defined by molecular heterogeneity across patients sharing the same diagnosis in a subject in need thereof, comprising: (a) performing (a) through (k) of claim 1; (b) compiling a list of the molecular targets of the disease; (c) aligning the molecular targets of the disease with the individualized PPI of the patient to find overlap between the disease targets and the patient PPI; (d) treating the patient with a treatment modality that overlaps with the patient’s individualized PPI 11 A method of claim 10 wherein the complex disease is a pulmonary vascular disease. 12 A method of claim 11 wherein the pulmonary vascular disease is pulmonary arterial hypertension.Attorney Docket No. 15024-398PC0 13. The method of claim 10 wherein the individualized PPI are patient- and time-specific wiring maps.

14. A method of claim 10 wherein the complex disease is selected from the group consisting of cardiomyopathy; parenchymal lung disease; solid tumors affecting the breast, lung, colon, pancreas, or ovaries; inflammatory bowel disease; systemic lupus erythematosus and systemic sclerosis; mixed disorders of cellular necrosis, apoptosis, cell proliferation, fibrosis, and metabolic dysfunction; and the like.

15. A method of claim 10 wherein the complex disease is Crohn’s disease or ulcerative colitis.

Images