Precision Medicine and Clinical Care¶
Chapter 5 | Part 1: The Profession of Medicine
KEY CLINICAL POINTS¶
- Precision medicine couples molecular reductionism with integrative, systems-level understanding of disease pathophysiology to individualize diagnosis and treatment
- Disease expression results from primary genetic/environmental drivers modified by an individual's unique genomic context (modifier genes), explaining phenotypic variability
- Convergent phenotypes (different diseases with similar presentations) and divergent phenotypes (same disease with different presentations) are both consequences of genomic context and environmental exposures
- Essential elements include deep phenotyping, endophenotyping, comprehensive genomic profiling, and understanding social determinants of health
- Network medicine approaches using protein-protein interaction networks (interactome) enable rational drug target identification and drug repurposing
1. DEFINITION & OVERVIEW¶
Precision medicine represents a paradigm shift from traditional clinicopathologic disease classification to an integrative approach that couples molecular reductionism with systems-level understanding of disease mechanisms. It aims to individualize diagnosis, prognosis, and treatment based on each patient's unique genomic context, environmental exposures, and phenotypic characteristics.
1.1 Evolution of Disease Nosology¶
Modern disease classification arose in the late nineteenth century, departing from holistic descriptions dating to Galen. The framework was institutionalized by Morgagni in 1761 with 'De Sedibus et Causis Morborum per Anatomen Indagatis,' correlating clinical features with over 600 autopsies. This clinicopathologic approach relied on inductive generalization and Occam's razor, reducing disease complexity to its simplest form. Diseases were characterized by end-organ manifestations and late-stage presentations.
1.2 Limitations of Traditional Approach¶
Oslerian diagnostics suffer from significant shortcomings including: failure to distinguish underlying etiology of different diseases with common pathophenotypes (e.g., many diseases cause end-stage kidney disease or heart failure); therapies may be ineffective due to lack of understanding of molecular complexities; upward of 60% of patients may not respond to the most 'effective' drugs; significant predictive inaccuracies affect large segments of disease populations.
1.3 The Arc of Reductionism¶
The historical progression of disease understanding illustrates the 'arc of reductionism': 18th century (sick person, phthisis) → Early 19th century (lesions of organs/tissues, caseating granulomata) → Late 19th century (lesions of cells/microbes, M. tuberculosis identification) → Late 20th century (molecular lesions, interferon testing) → 21st century (the challenge of reassembly). This arc revealed the need for new integrative approaches while respecting genomic context.
1.4 From Reductionism to Integration¶
The Human Genome Project provided new tools but disappointment followed as the pool of genomes expanded without expected revelations. Shortcomings are explained by the important roles of epigenetics, significantly modulated by environmental exposures and individual experiences. The modern approach is integrative, taking genomic context into account in all dimensions while remaining more useful than ancient holism.
2. FUNDAMENTAL PRINCIPLES¶
Two important confounding principles must be understood to develop a precision medicine strategy for any disease: convergent phenotypes and divergent phenotypes. Both are consequences of genomic context coupled with unique exposures over an individual's lifetime.
Examples of Convergent and Divergent Phenotypes¶
| Type | Clinical Example | Underlying Conditions/Presentations |
|---|---|---|
| Convergent | Hypertrophic myocardium | Hypertrophic cardiomyopathy (>11 sarcomeric proteins, >1400 variants), Hypertensive heart disease, Aortic stenosis, Fabry's disease, Pompe's disease |
| Convergent | Thrombotic microangiopathy | TTP, HUS, Malignant hypertension, Scleroderma renal crisis, Preeclampsia/eclampsia, HELLP syndrome, Antiphospholipid syndrome |
| Divergent | Aortic stenosis presentations | Syncope, Heart failure, Angina pectoris |
| Divergent | Antiphospholipid syndrome presentations | Venous thromboembolism, Thrombotic stroke, Mesenteric thrombosis, Coronary thrombosis, Livedo reticularis |
2.1 Convergent Phenotypes¶
Patients with different diseases can manifest similar pathophenotypes. Recognition of convergent phenotypes is essential for accurate diagnosis and appropriate treatment selection, as similar presentations may require vastly different therapeutic approaches based on underlying etiology.
2.2 Divergent Phenotypes¶
Patients with the same basic disease can manifest very different pathophenotypes. This includes different clinical manifestations of diseases like cystic fibrosis or sickle cell disease, and incomplete penetrance of many common genetic diseases. Understanding why some individuals with sickle cell anemia develop stroke while others develop acute chest syndrome exemplifies this principle.
2.3 Genomic Context and Modifier Genes¶
Modern genomics has established that genomic context (modifier genes) is distinctive for each individual. Understanding this context provides insight necessary to predict how primary disease drivers may manifest clinical pathophenotype. Primary genetic and/or environmental drivers differentially affect disease expression based on individual genomic context—the ultimate basis for precision medicine.
3. REQUIREMENTS FOR PRECISION MEDICINE¶
The essential elements of any precision medicine effort include phenotyping, endophenotyping (defining characteristics not readily observable), genomic profiling, and understanding social determinants of health. The level of precision needed is determined by actionability; otherwise, excessive testing leads to anxiety and accompanying risks.
Components of Comprehensive Precision Medicine Assessment¶
| Domain | Components | Clinical Application |
|---|---|---|
| Phenotyping | History, family history, environmental exposures, functional studies, imaging | Disease classification and subtyping |
| Endophenotyping | Biochemical tests, immunologic markers, molecular markers | Refining discriminant power |
| DNA Analysis | Exome sequencing, whole genome sequencing | Identifying disease drivers and variants |
| Gene Expression | mRNA profiling, protein expression | Understanding functional consequences |
| Metabolomics | Metabolite profiling | Pathway analysis and biomarker discovery |
| Epigenomics | DNA methylation, histone modifications | Environmental influence assessment |
| Proteomics | Post-translational modifications | Functional protein analysis |
| Metagenomics | Microbiome analysis | Host-microbe interactions |
| Immunophenotyping | Immune cell profiling, antibody patterns | Disease detection and monitoring |
3.1 Deep Phenotyping¶
Deep phenotyping requires: detailed history including family history and environmental exposures; relevant physiologic functional studies; imaging including molecular imaging where appropriate; biochemical, immunologic, and molecular tests of body fluids. Formalizing nuanced differences between individuals with the same disease is critical for achieving more precise phenotypes.
3.2 Endophenotyping¶
Objective laboratory tests together with functional studies compose an assessment of endophenotype (endotype), refining overall discriminant power of evaluation. Examples include distinctions among vasculitides facilitated by refinements in serologies or immunophenotyping. The impetus to create disease subclasses is largely determined by need to improve prognosis and apply more precise therapies.
3.3 Orthogonal Phenotyping¶
Orthogonal phenotyping assesses clinical, molecular, imaging, or functional endophenotypes seemingly unrelated to clinical presentation. These features enhance ability to distinguish subphenotypes, deriving from the fact that diseases can be subtly (subclinically) manifest in organ systems different from that in which primary symptoms or signs are expressed. Unbiased comprehensive phenotyping should become the norm.
3.4 Genomic Profiling Components¶
Complex levels of genomic assessment include: DNA sequencing (exome, whole genome); gene expression (mRNA and protein expression); metabolomics; epigenome; posttranslationally modified proteome; metagenome (personal microbiome). Immunophenotyping is an emerging area using the immune system as an indicator of disease, prior exposures, and sensor for emergence of new diseases.
3.5 Tissue-Specific Considerations¶
While DNA sequencing using whole blood generally applies to any organ-based disease, gene expression, metabolomics, and epigenomics are often tissue-specific. Attempts at correlating whole-blood mRNA, protein, or metabolite profiles with those of involved organs are critical for precise prognostics and therapeutic choices. A key determinant of functional consequences of genetic variants is not simply expression in tissue of interest, but coexpression of protein binding partners comprising specific (dys)functional pathways.
3.6 Single-Cell Gene Expression¶
Single-cell gene expression data add complexity to understanding disease phenotype genesis. These data are becoming increasingly available for different cell and tissue types, including spatial distribution. The role of differential expression on ultimate integrative pathophenotype and how intercellular communication may influence or be influenced by differential gene expression remain questions of ongoing study.
3.7 Temporal Considerations¶
While phenotype features of many chronic diseases are assessed longitudinally, genomic features tend to be limited to single time point sampling. Time trajectories are extremely informative in precision genotyping and phenotyping, with gene expression patterns and phenotypes changing over time differently among patients with the same overarching phenotype. Cost, feasible sampling frequency, predictive power, and therapeutic choices will drive optimal strategy for timed sample acquisition.
4. NETWORK MEDICINE APPROACH¶
The protein-protein interaction network (interactome) serves as a comprehensive network map of protein-protein interactions in a cell or organ of interest. This template provides information on subnetworks that govern disease phenotype (disease modules), which can be individualized by incorporating patient-specific variants and differentially expressed proteins.
Network Medicine Framework¶
| Component | Description | Application |
|---|---|---|
| Human Interactome | Complete protein-protein interaction network | Template for disease module identification |
| Disease Module | Subnetwork governing specific disease phenotype | Target identification, pathway analysis |
| Tissue-Specific Network | Subgraph of significantly expressed genes in tissue | Understanding organ-specific disease manifestations |
| Reticulotype | Individual-specific network linking genotype to phenotype | Personalized therapeutic targeting |
| Network Proximity | Distance metrics between drug targets and disease genes | Drug repurposing opportunities |
4.1 The Interactome¶
The human interactome represents a comprehensive map of protein-protein interactions. Current data includes approximately 13,460 proteins and 141,296 interactions across 70 diseases and 64 tissues. This network serves as the template for understanding how disease genes interact and form functional modules.
4.2 Disease Modules¶
Disease modules are subnetworks within the interactome that govern specific disease phenotypes. Identifying these modules allows understanding of: tissue specificity of disease gene expression; reduction of total disease module genes governing phenotype in specific organs; specific pathways expressed in functional entirety in particular tissues.
4.3 Reticulotype Concept¶
The reticulotype (from Latin for network) links genotype to phenotype of an individual. It is created by integrating: patient-specific molecular perturbations; genetic variants; differentially expressed genes; unique integrative biologic network context. This individualized network structure enables patient-specific precision therapeutics.
4.4 Network-Based Drug Target Identification¶
Using the interactome approach, one can identify potential drug targets rationally by examining disease modules and their component proteins. Drug targets are evaluated based on their position within disease modules and their ability to modulate disease-relevant pathways.
4.5 Network-Based Drug Repurposing¶
The proximity hypothesis enables drug repurposing by demonstrating proximity of known drug targets to disease modules of interest. Existing drugs approved for other purposes can be evaluated for new indications based on their network proximity to newly identified disease targets. Example: In multicentric Castleman's disease, recognition of PI3K/Akt/mTOR pathway activation led to sirolimus trials.
5. CANCER AND PRECISION MEDICINE¶
Cancer represents a prime example of precision medicine opportunity where tumors can be sampled frequently to monitor temporal changes in the somatically mutating oncogenome. There are a limited number of oncogenic pathways (<20) represented in the majority of malignancies, regardless of organ of origin.
Precision Therapeutics in Cancer¶
| Drug | Target | Disease |
|---|---|---|
| Imatinib | Bcr-Abl tyrosine kinase | Chronic myeloid leukemia |
| Erlotinib | EGFR mutations | Non-small cell lung cancer |
| Ibrutinib | Bruton tyrosine kinase | Chronic lymphocytic leukemia |
5.1 Genomic Profiling in Cancer¶
Cancers can be frequently sampled (biopsied) to monitor temporal changes in the oncogenome and consequences for oncogenic driver pathways. Genomic signatures serve as templates for precisely targeted therapies leading to dramatic changes in treatment response.
5.2 Examples of Precision Cancer Therapeutics¶
Successful examples include: Imatinib (and congeners) for Bcr-Abl tyrosine kinase activity in chronic myeloid leukemia; Erlotinib for EGFR-mutant non-small cell lung cancers; Ibrutinib for Bruton tyrosine kinase in chronic lymphocytic leukemia.
5.3 Challenges Unique to Cancer Precision Medicine¶
Four primary challenges: (1) Mutational landscape evolves as disease progresses; therapy leads to selection for resistant clones; (2) Likelihood of cure by single agent is limited, necessitating rational polypharmaceutical approaches considering alternative pathways; (3) Marked genomic heterogeneity within tumors argues that targeting specific pathways may not succeed long-term due to continued heterogeneous genomic evolution; (4) Variability in patient characteristics and ability to withstand treatment and mount complementary immune response.
5.4 Tumor Microenvironment Considerations¶
In solid tumors, stromal cells interact with malignant cells in various ways (e.g., metabolically), and their gene expression signatures are modified by the changing somatic mutational landscape. Single-cell mRNA sequencing demonstrates great variability between apparently similar cells, challenging functional interpretation.
6. CLINICAL APPLICATIONS BY SPECIALTY¶
Precision medicine applications span the pregenomic and postgenomic eras, with progressively refined methods characterizing pathophenotypes and genotypes matched to idealized therapeutic combinations.
Precision Medicine Applications by Specialty¶
| Specialty | Condition | Precision Approach | Key Finding/Application |
|---|---|---|---|
| Cardiology | Heart Failure | Phenotype-guided therapy | HFrEF vs HFpEF distinction guides therapy |
| Pulmonology | Pulmonary Arterial Hypertension | Genotype-phenotype correlation | NEDD-9 fibrotic endophenotype identification |
| Neurology | Frontotemporal Dementia | Genetic classification | Specific genotypes linked to clinical presentations |
| Neurology | Neuromyelitis Optica | Autoantibody profiling | AQP-4 and MOG antibodies distinguish from MS |
| Neurology | Myasthenia Gravis | Autoantibody stratification | Novel antibodies enable precision therapy |
| Pharmacology | Drug Metabolism | Pharmacogenomic testing | TPMT, CYP2C19 variant testing |
6.1 Cardiovascular Disease¶
Heart Failure: Pregenomic precision medicine uses diuretics, digoxin, beta blockers, afterload-reducing agents, venodilators, renin-angiotensin-aldosterone inhibitors, and nesiritide in combination tailored to primary pathophysiologic phenotypes (congestion, hypertension, impaired contractility). Subclassification into HFrEF and HFpEF based on echocardiographic assessments, with recognition that HFpEF responds poorly to most current therapeutic classes.
6.2 Pulmonary Vascular Disease¶
Pulmonary Arterial Hypertension: Prior to 1990s, no effective therapies existed. Molecular and biochemical characterization led to therapies restoring normal vascular function: calcium channel blockers, prostacyclin congeners, endothelin receptor antagonists. Genomic characterization reveals distinct genotypes yielding unique phenotypes, such as fibrotic endophenotype governed by oxidized NEDD-9 and aldosterone-dependent, TGF- β –independent enhancement of collagen III expression.
6.3 Endocrine Disease¶
Type 1 Diabetes Mellitus: Treated with goal of restoring metabolic normality, precisely titrating hormone treatments (insulin) to metabolic endpoint (glucose). Represents pregenomic precision medicine with individualized dosing based on metabolic response.
6.4 Neurological Disease - Dementias¶
Precision diagnostics has led to new dementia classification based on genes and pathways involved and sites where aggregated proteins first form and spread. Frontotemporal dementia presentations (progressive aphasia, behavioral disturbances, dementia with ALS) can now be linked to specific genotypes and susceptible cell types. Prion diseases: clinical phenotype determined by specific germline mutations in prion protein.
6.5 Neurological Disease - Autoimmune¶
Neuromyelitis Optica: Discovery of autoantibodies against aquaporin-4 (AQP-4) and myelin oligodendrocyte glycoprotein (MOG) allowed classification as separate entity from multiple sclerosis, requiring different treatment. Myasthenia Gravis: Identification of novel autoantibodies permits stratification and more finely tuned precision approach to therapy.
6.6 Pharmacogenomics¶
Precision medicine optimizes drug utilization by assessing individualized pharmacogenomics of disposition and metabolism. Key examples: TPMT variants affecting azathioprine metabolism; CYP2C19 variants affecting clopidogrel metabolism.
7. INFECTIOUS DISEASE PARALLELS¶
Precision approaches in infectious disease parallel cancer strategies through identification of causative organisms and sensitivity to antimicrobials. Combinatorial antimicrobial treatments address acquired resistance.
7.1 Organism Identification and Sensitivity¶
Identification of causative organism and antimicrobial sensitivity allows precision treatment approaches. These diagnostic and therapeutic strategies can be applied without detailed knowledge of personalized responses to infection or treatment (aside from serious adverse effects) with good outcomes in most cases.
7.2 Individual Response Variability¶
Individuals respond differently to specific infections and treatments, possibly driven by different endophenotypes (e.g., different inflammatory responses). More precise knowledge of these mechanistic differences may yield improved prognosis and therapeutic approaches.
7.3 Immune Modulation¶
Immune modulation, particularly for immune exhaustion in chronic infections, represents a new frontier amenable to personalized, precise analyses. This parallels approaches in cancer therapy.
8. DATA REQUIREMENTS AND INFORMATICS¶
Six dimensions characterize individuals in the precision medicine era: health system data, 'omic' data, study-participant-generated data, exposome/social determinants, motivations and behaviors, and microbiome. Integration of these dimensions creates the precision participant descriptor.
Six Dimensions of Precision Medicine Characterization¶
| Dimension | Components | Purpose |
|---|---|---|
| Health System Data | EMR, clinical encounters, diagnoses, treatments | Clinical phenotyping |
| 'Omic' Data | Genomic, transcriptomic, proteomic, metabolomic | Molecular profiling |
| Participant-Generated Data | Wearables, patient-reported outcomes | Real-world phenotyping |
| Exposome/Social Determinants | Environmental exposures, social history | Contextual factors |
| Motivations and Behaviors | Lifestyle, adherence patterns | Behavioral phenotyping |
| Microbiome | Gut and other microbiome compositions | Host-microbe interactions |
8.1 Big Data in Precision Medicine¶
Six dimensions of individual characterization: (1) Health system data; (2) 'Omic' data; (3) Study-participant-generated data; (4) Exposome/social determinants; (5) Motivations and behaviors; (6) Microbiome. The precision participant descriptor integrates data from these dimensions and varies over time.
8.2 Data Storage and Analysis¶
Large phenotypic and genomic data sets require sufficient storage for analysis, especially for individuals with time trajectories (which should be acquired whenever feasible). Analytical methods required to extract useful information are complex and evolving.
8.3 Electronic Health Record Evolution¶
The electronic medical record must evolve to provide curated precision data in user-friendly display. Medical informatics efforts must harmonize data sets, standardize data collection, and optimize/standardize data analysis.
8.4 Machine Learning and AI¶
Machine learning and artificial (auxiliary) intelligence methods are essential for extracting optimal information from data sets. These include pathways that can be uniquely targeted therapeutically and individualized genomic or phenotypic signatures highly predictive of outcome.
9. CLINICAL TRIAL DESIGN AND ENRICHMENT¶
Precision medicine evolved partly from clinical trial design. From the entire population with disease of interest, a sample cohort is enrolled that ideally represents the entire distribution. Enrichment strategies facilitate trial conduct and improve precision in defining treatment response.
Clinical Trial Enrichment Strategies¶
| Strategy | Approach | Advantage | Limitation |
|---|---|---|---|
| Decreased Heterogeneity | Narrow inclusion criteria | Enhanced effect size | May miss responders outside criteria |
| Prognostic Enrichment | Select high-risk patients | More events, smaller sample size | Does not predict treatment response |
| Predictive Enrichment | Use biomarkers/characteristics predicting response | Identifies likely responders | Requires validated predictive markers |
9.1 Decreased Heterogeneity Strategy¶
Decreasing heterogeneity in study population enhances effect size but is based on analyses of prior data sets defining individuals more likely to respond to therapy. This approach improves trial conduct but may not improve precision for individual patients.
9.2 Prognostic Enrichment¶
Increasing representation of individuals with high risk of observed outcomes facilitates trial conduct. Selects for patients more likely to experience endpoints but does not necessarily predict treatment response.
9.3 Predictive Enrichment¶
Utilizes both trial participant characteristics and data from experiments conducted before or during (adaptive design) the trial. Improves prediction of who is likely to have more pronounced response to treatment under study. Follows from information provided by detailed, big data-driven analysis exploring phenotypic and genomic features used to predict response.
9.4 Practical Boundaries of Precision¶
Features defining responders need not be precisely met by each patient; they can be collated or clustered to define reasonably sized cohort predicted to respond within certain confidence bounds. Boundaries to precision medicine practice are imprecise strictly speaking, but sufficiently predictive to be practical from perspectives of clinical care and cost-effectiveness.
10. FUTURE DIRECTIONS AND CHALLENGES¶
Precision medicine holds great promise for the future practice of medicine, but several requirements must be met for continued successful evolution.
10.1 Data Requirements¶
Deeply refined personal phenotypic data and genomic data are essential. While great progress has been made in genomics and biochemical testing, ability to capture meaningful immunologic endophenotypes and environmental exposures is limited. Normal population comparison data sets are required for optimal prediction.
10.2 Phenotyping Expansion¶
Phenotyping must continue to expand and become dimensionally richer. Must incorporate: data relevant to clinical presentation; orthogonal phenotypic data yielding useful information on disease trajectory or preclinical disease markers; personal device data; environmental exposure history; social network interactions; health system data.
10.3 Minimal Data Set Challenge¶
The greatest challenge is determining minimal data set required to predict outcome and response to therapy. Goals include: eliminating redundant information in overdetermined biologic systems; weighting determinants of outcome; using data as phenomic/genomic signatures easier to collect than comprehensive unbiased data sets.
10.4 Analytical Methods¶
Rapidly evolving machine learning and artificial intelligence strategies are essential for maximal success. These methods will extract useful information including pathways for therapeutic targeting and individualized signatures predictive of outcome.
11. KEY POINTS & CLINICAL PEARLS¶
Critical concepts for clinical practice of precision medicine.
Clinical Pearls in Precision Medicine¶
| Principle | Clinical Implication |
|---|---|
| Convergent phenotypes | Similar presentations may have vastly different etiologies requiring different treatments |
| Divergent phenotypes | Same disease can manifest differently; look for underlying genomic explanations |
| Genomic context matters | Modifier genes explain why same mutation has different clinical effects in different patients |
| Orthogonal phenotyping | Assess organ systems beyond the primary clinical presentation |
| Network proximity | Existing drugs may be repurposed based on target proximity to disease modules |
| Time trajectories | Gene expression and phenotypes change over time—serial assessment is valuable |
| Actionable precision | More precision is needed when it is actionable; otherwise leads to excessive testing |
| Tissue specificity | Gene expression consequences depend on coexpression of binding partners in specific tissues |
| Cancer evolution | Mutational landscape evolves with treatment; plan for resistance mechanisms |
| Practical boundaries | Precision medicine can be practically applied without complete genomic knowledge |