What is personalized medicine and what should it replace?

Abstract

Personalized medicine is a new framework for medical care that involves modelling and simulation of a disease on the basis of its underlying mechanisms. This strategy must replace the 20(th) century paradigm of defining disease by pathology or associated signs and symptoms and conducting outcomes research that is based on the presence or absence of the disease syndrome. New technologies, including next-generation sequencing, the 'omics' and powerful computers provide massive amounts of accurate data. However, attempts to understand complex disorders by applying these new technologies within the 20(th) century framework have failed to produce the expected medical advances. To help physicians embrace a paradigm shift, the limitations of the old framework and major advantages of the new framework must be demonstrated. Chronic pancreatitis is an ideal complex disorder to study to consider the pros and cons of the two frameworks, because the pancreas is such a simple organ for disease modelling, and the advantages of personalized medicine are so profound.

Figure 1

Two approaches to assess individual risk on the basis of the presence or absence of six variables. a | Traditional subclassification approaches divide the population into 32 categories, record the outcomes and then estimate the individual’s risk based on the category with characteristics most like the individual. b | Modelling and simulation approaches can start with results from subclassification approaches and improve on them using statistical and mechanistic (deterministic) approaches based on knowledge of biology. Risk estimates for independent or combined variables can be calculated to replicate the results of the population study and provide new insights (for example, Archimedes). Modelling is useful for simulation population studies, and also for an individual patient. For an individual patient, modelling provides prognostic information that can be more accurate when it uses continuous variables rather than arbitrary cut-off values, when the patient is in a poorly-populated subclassification with limited outcomes data, when deterministic relationships based on biological mechanisms are included, when new variables of known mechanistic effect are added (for example, genetic risk) or when their disease falls outside of current recommendations and population studies to determine complex risk subcategories cannot be done owing to small patient numbers or high costs. Rounded rectangles, patient populations; diamonds, decision points; hexagon, risk calculator; DM, diabetes mellitus; LDL, low density lipoproteins; SBP, systolic blood pressure; smoke, history of smoking; SC, subclassification.

Figure 2

Clinical pathway of patients being evaluated for suspected pancreatic disease. a | Referred patients undergo initial evaluation and triage. b | Patients with idiopathic or complicated pancreatic diseases undergo genetic testing and risk modelling. c | Patients are classified into general disease mechanism groups and treatment effectiveness is monitored. d | Common complications are also anticipated and addressed, with appropriate reference to sequencing data results. Patients with unexpected outcomes are re-evaluated and additional assessment is done as needed. Abbreviations: AP, acute pancreatitis; CF, cystic fibrosis; CP, chronic pancreatitis; dz, disease; ERCP, endoscopic retrograde cholangiopancreatography; NGS, next-generation sequencing; PERT, pancreatic enzyme replacement therapy; PC, pancreatic cancer; RAP, recurrent acute pancreatitis; Rx, treatment; TP/IAT, Total pancreatectomy with islet autotransplantation.