Operational implementation of prospective genotyping for personalized medicine: the design of the Vanderbilt PREDICT project

The promise of "personalized medicine" guided by an understanding of each individual's genome has been fostered by increasingly powerful and economical methods to acquire clinically relevant information. We describe the operational implementation of prospective genotyping linked to an advanced clinical decision-support system to guide individualized health care in a large academic health center. This approach to personalized medicine entails engagement between patient and health-care provider, identification of relevant genetic variations for implementation, assay reliability, point-of-care decision support, and necessary institutional investments. In one year, approximately 3,000 patients, most of whom were scheduled for cardiac catheterization, were genotyped on a multiplexed platform that included genotyping for CYP2C19 variants that modulate response to the widely used antiplatelet drug clopidogrel. These data are deposited into the electronic medical record (EMR), and point-of-care decision support is deployed when clopidogrel is prescribed for those with variant genotypes. The establishment of programs such as this is a first step toward implementing and evaluating strategies for personalized medicine.

Predicting clopidogrel response using DNA samples linked to an electronic health record

Variants in ABCB1 and CYP2C19 have been identified as predictors of cardiac events during clopidogrel therapy initiated after myocardial infarction (MI) or percutaneous coronary intervention (PCI). In addition, PON1 has recently been associated with stent thrombosis. The reported effects of these variants have not yet been replicated in a real-world setting. We used BioVU, the Vanderbilt DNA repository linked to de-identified electronic health records (EHRs), to find data on patients who were on clopidogrel treatment after an MI and/or a PCI; among these, we identified those who had experienced one or more recurrent cardiac events while on treatment (cases, n = 225) and those who had not experienced any cardiac event while on treatment (controls, n = 468). We found that CYP2C19*2 (hazard ratio (HR) 1.54, 95% confidence interval (CI) 1.16-2.06, P = 0.003) and ABCB1 (HR 1.28, 95% CI 1.04-1.57, P = 0.018), but not PON1 (HR 0.91, 95% CI 0.73-1.12, P = 0.370), were associated with recurrent events. In this population, genetic signals for clopidogrel resistance in ABCB1 and CYP2C19 were replicated, supporting the use of EHRs for pharmacogenomic studies. Our data do not show an association between PON1 and recurrent cardiovascular events.

Pharmacogenomics: application to the management of cardiovascular disease

The past decade has seen substantial advances in cardiovascular pharmacogenomics. Genetic determinants of response to clopidogrel and warfarin have been defined, resulting in changes to the product labels for these drugs that suggest the use of genetic information as a guide for therapy. Genetic tests are available, as are guidelines for incorporation of genetic information into patient-care decisions. These guidelines and the literature supporting them are reviewed herein. Significant advances have also been made in the pharmacogenomics of statin-induced myopathy and the response to β-blockers in heart failure, although the clinical applications of these findings are less clear. Other areas hold promise, including the pharmacogenomics of antihypertensive drugs, aspirin, and drug-induced long-QT syndrome (diLQTS). The potential value of pharmacogenomics in the discovery and development of new drugs is also described. In summary, pharmacogenomics has current applications in the management of cardiovascular disease, with clinically relevant data continuing to mount.

The emerging role of electronic medical records in pharmacogenomics

Health-care information technology and genotyping technology are both advancing rapidly, creating new opportunities for medical and scientific discovery. The convergence of these two technologies is now facilitating genetic association studies of unprecedented size within the context of routine clinical care. As a result, the medical community will soon be presented with a number of novel opportunities to bring functional genomics to the bedside in the area of pharmacotherapy. By linking biological material to comprehensive medical records, large multi-institutional biobanks are now poised to advance the field of pharmacogenomics through three distinct mechanisms: (i) retrospective assessment of previously known findings in a clinical practice-based setting, (ii) discovery of new associations in huge observational cohorts, and (iii) prospective application in a setting capable of providing real-time decision support. This review explores each of these translational mechanisms within a historical framework.

Estimation of the warfarin dose with clinical and pharmacogenetic data

BACKGROUND

Genetic variability among patients plays an important role in determining the dose of warfarin that should be used when oral anticoagulation is initiated, but practical methods of using genetic information have not been evaluated in a diverse and large population. We developed and used an algorithm for estimating the appropriate warfarin dose that is based on both clinical and genetic data from a broad population base.

METHODS

Clinical and genetic data from 4043 patients were used to create a dose algorithm that was based on clinical variables only and an algorithm in which genetic information was added to the clinical variables. In a validation cohort of 1009 subjects, we evaluated the potential clinical value of each algorithm by calculating the percentage of patients whose predicted dose of warfarin was within 20% of the actual stable therapeutic dose; we also evaluated other clinically relevant indicators.

RESULTS

In the validation cohort, the pharmacogenetic algorithm accurately identified larger proportions of patients who required 21 mg of warfarin or less per week and of those who required 49 mg or more per week to achieve the target international normalized ratio than did the clinical algorithm (49.4% vs. 33.3%, P<0.001, among patients requiring < or = 21 mg per week; and 24.8% vs. 7.2%, P<0.001, among those requiring > or = 49 mg per week).

CONCLUSIONS

The use of a pharmacogenetic algorithm for estimating the appropriate initial dose of warfarin produces recommendations that are significantly closer to the required stable therapeutic dose than those derived from a clinical algorithm or a fixed-dose approach. The greatest benefits were observed in the 46.2% of the population that required 21 mg or less of warfarin per week or 49 mg or more per week for therapeutic anticoagulation.

Influence of the G2677T/C3435T haplotype of MDR1 on P-glycoprotein trafficking and ibutilide-induced block of HERG

The drug efflux pump P-glycoprotein possesses two common and often linked polymorphisms that result in variable drug action. G2677T results in A893S, whereas C3435T is synonymous and has been reported to alter protein folding. We tested the effect of these MDR1 variants on Human Ether-Related A Go-Go (HERG) block by ibutilide in CHO cells 48 h following transient transfection with an IRES-dsRed vector containing MDR1, G2677T MDR1, G2677T/C3435T MDR1 or an empty bicistronic site and an IRES-GFP vector containing HERG (KCNH2). Cotransfection of MDR1 variants had no effect on I(Kr) amplitude at baseline. Cells cotransfected with MDR1-G2677T showed resistance to ibutilide vs HERG alone (IC(50): 105.3+/-1.42 nM vs 27.4+/-2.5 nM; P<0.0001), consistent with the idea that A893S attenuates I(Kr) block by enhancing drug efflux and thus reducing the drug available to interact with the channel binding site. However, G2677T/C3435T cells showed ibutilide sensitivity similar to cells expressing HERG alone (IC(50): 22.2+/-0.9 nM). Immunostaining showed that the C3435T variant did not traffic to the cell surface. Coculture with fexofenadine(1 microM), an MDR1 substrate known to rescue misfolding in other membrane proteins, restored cell surface expression of MDR1 G2677T/C3435T and restored resistance to block HERG by ibutilide 200 nM (98.5+/-0.98% vs 42.3+/-2.2%, P<0.001). The non-synonymous MDR1 variant G2677 T (A893S) confers resistance to ibutilide block of I(Kr), which is mitigated by the C3435T polymorphism through reduced protein expression, an effect that can be restored by coculture with fexofenadine. These data identify ibutilide as an MDR1 substrate and further support the concept that variable drug transport function can modulate the action of HERG blockers.

A practical approach to torsade de pointes

The term torsade de pointes refers to polymorphic ventricular tachycardia that occurs in the setting of an abnormally long QT interval. While the most common cause is treatment with QT prolonging drugs, torsade de pointes also occurs in the congenital long QT syndromes and in the setting of acquired heart block or severe electrolyte disturbance, notably hypokalemia. Among QT prolonging drugs that cause torsade de pointes, both antiarrhythmics and "noncardioactive" drugs have been recognized. The electrocardiographic features of torsade de pointes include labile QT intervals, prominent U waves, and a "pause-dependent" onset of the arrhythmia. Treatment consists of recognition of the syndrome, correction of underlying electrolyte abnormalities, and withdrawal of any offending drugs. Magnesium, isoproterenol, or cardiac pacing provides specific antiarrhythmic therapy in torsade de pointes.

Polymorphisms in the cardiac sodium channel promoter displaying variant in vitro expression activity

Variable transcription of the cardiac sodium channel gene is a candidate mechanism determining arrhythmia susceptibility. We have previously cloned and characterized the core promoter and flanking region of SCN5A, encoding the cardiac sodium channel. Loss-of-function mutations in this gene have been reported in approximately 20% of patients with Brugada syndrome, an inherited cardiac electrical disorder associated with a high incidence of life-threatening arrhythmias. In this study, we identified DNA variants in the proximal 2.8 kb promoter region of SCN5A and determined their frequency in 1,121 subjects. This population consisted of 88 Brugada syndrome patients with no SCN5A coding region mutation, and 1,033 anonymized subjects from various ethnicities. Variant promoter activity was assayed in CHO cells and neonatal cardiomyocytes by transient transfection of promoter-reporter constructs. Single-nucleotide polymorphisms (SNPs) were identified at approximately 1/200 base pairs which are: 11 in the 5'-flanking region, 1 in exon 1, and 5 in intron 1. In addition, a haplotype consisting of two SNPs in complete linkage disequilibrium was identified. Minor allele frequencies were >5% in at least one ethnic panel at 5/19 polymorphic sites. In vitro functional analysis in cardiomyocytes identified four variants with significantly (P<0.05) reduced reporter activity (up to 63% reduction). The largest changes were seen with c.-225-1790 G>A, which reduced reporter activity by 62.8% in CHO cells and 55% in cardiomyocytes. From these results, we can conclude that the SCN5A core promoter includes multiple DNA polymorphisms with altered in vitro activity, further supporting the concept of interindividual variability in transcription of this cardiac ion channel gene.

The pharmacogenetics research network: from SNP discovery to clinical drug response

The NIH Pharmacogenetics Research Network (PGRN) is a collaborative group of investigators with a wide range of research interests, but all attempting to correlate drug response with genetic variation. Several research groups concentrate on drugs used to treat specific medical disorders (asthma, depression, cardiovascular disease, addiction of nicotine, and cancer), whereas others are focused on specific groups of proteins that interact with drugs (membrane transporters and phase II drug-metabolizing enzymes). The diverse scientific information is stored and annotated in a publicly accessible knowledge base, the Pharmacogenetics and Pharmacogenomics Knowledge base (PharmGKB). This report highlights selected achievements and scientific approaches as well as hypotheses about future directions of each of the groups within the PGRN. Seven major topics are included: informatics (PharmGKB), cardiovascular, pulmonary, addiction, cancer, transport, and metabolism.

Proarrhythmia

The concept that antiarrhythmic drugs can exacerbate the cardiac rhythm disturbance being treated, or generate entirely new clinical arrhythmia syndromes, is not new. Abnormal cardiac rhythms due to digitalis or quinidine have been recognized for decades. This phenomenon, termed "proarrhythmia," was generally viewed as a clinical curiosity, since it was thought to be rare and unpredictable. However, the past 20 years have seen the recognition that proarrhythmia is more common than previously appreciated in certain populations, and can in fact lead to substantially increased mortality during long-term antiarrhythmic therapy. These findings, in turn, have moved proarrhythmia from a clinical curiosity to the centerpiece of antiarrhythmic drug pharmacology in at least two important respects. First, clinicians now select antiarrhythmic drug therapy in a particular patient not simply to maximize efficacy, but very frequently to minimize the likelihood of proarrhythmia. Second, avoiding proarrhythmia has become a key element of contemporary new antiarrhythmic drug development. Further, recognition of the magnitude of the problem has led to important advances in understanding basic mechanisms. While the phenomenon of proarrhythmia remains unpredictable in an individual patient, it can no longer be viewed as "idiosyncratic." Rather, gradations of risk can be assigned based on the current understanding of mechanisms, and these will doubtless improve with ongoing research at the genetic, molecular, cellular, whole heart, and clinical levels.

Long QT syndrome: reduced repolarization reserve and the genetic link

Marked QT prolongation and torsades de pointes can occur not only in the congenital long QT syndromes (LQTSs) but also as a consequence of environmental stimuli, notably administration of certain drugs. A key feature of this 'acquired' form of the LQTS has been its unpredictable nature. That is, although risk factors have been identified in series of patients, they have not been terribly useful in addressing risk in an individual patient. Normal cardiac repolarization depends critically on the interplay of multiple ion currents, and these provide some redundancy, or 'reserve', to protect against excessive QT prolongation by drugs. We have proposed that lesions in these repolarizing mechanisms can remain subclinical but nevertheless increase risk on drug exposure, and have termed this situation 'reduced repolarization reserve'. The evidence in support of this concept is presented, and the known and potential contributions by genetic variants to risk is examined. Assessing variability in susceptibility to acquired LQTS provides a framework for analysis of other complex gene-environment interactions.

Drug block of I(kr): model systems and relevance to human arrhythmias

The long QT-related arrhythmia torsades de pointes (TdP) can arise with mutations in HERG and during treatment with drugs that block cardiac I Kr, the current encoded by HERG. Multiple test systems have been used to assess drug block of I Kr. This study evaluated the I Kr blocking potency of a series of antiarrhythmics associated with a range of clinical risks of TdP in two such systems: mouse AT-1 cells (in which I Kr is the major repolarizing current) and Ltk cells transiently transfected with HERG (n = 4-10 cells per drug). For each compound, the concentration required to produce 50% block of I Kr or HERG tail currents (IC 50 ) was determined. There was an excellent correlation ( r = 0.98, p < 10 -5 ) between values obtained in the two systems. However, the relation between the liability of a drug to cause TdP appeared dissociated from I Kr blocking potency. Quinidine, dofetilide, ibutilide, procainamide, and disopyramide are all associated with TdP, but only the first three were potent blockers (IC 50 < or = 1 microM ), whereas procainamide and disopyramide were not (IC 50 > 50 microM ). Conversely, verapamil and amiodarone, drugs not associated with TdP, were also blockers (IC 50 < or = 1 microM ). We conclude that I Kr blocking potency can be readily assessed in either AT-1 cells or systems in which HERG is heterologously expressed. However, not all drugs causing TdP are potent I Kr blockers, and I Kr block is not necessarily associated with TdP. Other properties of these drugs, therefore, contribute to their propensity to cause TdP.

Gating-dependent mechanisms for flecainide action in SCN5A-linked arrhythmia syndromes

BACKGROUND

Mutations in the cardiac sodium (Na) channel gene (SCN5A) give rise to the congenital long-QT syndrome (LQT3) and the Brugada syndrome. Na channel blockade by antiarrhythmic drugs improves the QT interval prolongation in LQT3 but worsens the Brugada syndrome ST-segment elevation. Although Na channel blockade has been proposed as a treatment for LQT3, flecainide also evokes "Brugada-like" ST-segment elevation in LQT3 patients. Here, we examine how Na channel inactivation gating defects in LQT3 and Brugada syndrome elicit proarrhythmic sensitivity to flecainide.

METHODS AND RESULTS

We measured whole-cell Na current (I(Na)) from tsA-201 cells transfected with DeltaKPQ, a LQT3 mutation, and 1795insD, a mutation that provokes both the LQT3 and Brugada syndromes. The 1795insD and DeltaKPQ channels both exhibited modified inactivation gating (from the closed state), thus potentiating tonic I(Na) block. Flecainide (1 micromol/L) tonic block was only 16.8+/-3.0% for wild type but was 58.0+/-6.0% for 1795insD (P<0.01) and 39.4+/-8.0% (P<0.05) for DeltaKPQ. In addition, the 1795insD mutation delayed recovery from inactivation by enhancing intermediate inactivation, with a 4-fold delay in recovery from use-dependent flecainide block.

CONCLUSIONS

We have linked 2 inactivation gating defects ("closed-state" fast inactivation and intermediate inactivation) to flecainide sensitivity in patients carrying LQT3 and Brugada syndrome mutations. These results provide a mechanistic rationale for predicting proarrhythmic sensitivity to flecainide based on the identification of specific SCN5A inactivation gating defects.

Pharmacogenetics and drug-induced arrhythmias

Drugs are widely recognized to vary in the beneficial and undesirable effects they produce in human subjects. The understanding that variants (polymorphisms and mutations) in the human genome are common and may well modulate both disease and its response to drugs, is a critical new concept in understanding mechanisms of drug action and their variability in human subjects. Variability can arise because of variability in genes encoding molecules of drug disposition, in genes encoding molecules that drugs target, or in genes that modulate the overall activity of the complex biological systems within which drugs act. The evolving understanding of the genetic basis of variability in response to drugs used in the treatment of sudden cardiac death has important implications not only for the treatment of patients who have survived an episode, but also for helping formulate a framework for further understanding mechanisms of drug action at the genetic level.

Pause-dependent polymorphic ventricular tachycardia during long-term treatment with dofetilide: a placebo-controlled, implantable cardioverter-defibrillator-based evaluation

OBJECTIVES

To compare the incidence of pause-dependent polymorphic ventricular tachycardia (PVT) in patients with implantable cardioverter-defibrillators (ICDs) randomly assigned to the QT-prolonging antiarrhythmic dofetilide or placebo.

BACKGROUND

Drug-related torsade de pointes (TdP) is usually recognized within days of initiating therapy, but its incidence during long-term therapy is unknown.

METHODS

We assessed the frequency of TdP and ICD electrograms compatible with TdP in a multicenter study that randomized ICD patients to placebo (n = 87) or dofetilide (n = 87). As reported elsewhere, the number of patients with a primary trial end point (ICD intervention for VT or ventricular fibrillation) was similar in the two groups. For this analysis, a qualifying event was TdP (on electrocardiogram) or an intracardiac electrogram showing pause-dependent PVT.

RESULTS

A total of 620 electrograms obtained in 131 patients were analyzed blindly by prospectively defined criteria for episodes of pause-dependent polymorphic VT. These were identified in 15/87 (17%) patients receiving dofetilide and 5/87 (6%) patients on placebo (p < 0.05). Five of these episodes were early (<3 days), all of which were TdP on dofetilide. There were 15 late events, 10 on dofetilide and five on placebo (p = 0.29). The median time to a late event was 22 days (range 6 to 107 days) for dofetilide and 99 days (range 34 to 207 days) for placebo.

CONCLUSIONS

Pause-dependent PVT was more common among patients receiving dofetilide, although total VT incidence was similar in the two groups. These data suggest that in ICD patients either long-term dofetilide therapy is associated with an increased risk of TdP or the drug alters VT morphology.

Potassium current antagonist properties and proarrhythmic consequences of quinolone antibiotics

Quinolones are clinically important antibiotic drugs. One quinolone antibiotic, sparfloxacin (SPX), has been recently reported to increase the QT interval, and another quinolone, grepafloxacin (GRX), was withdrawn because it induced torsade de pointes (TdP), a polymorphic ventricular tachycardia (VT) linked to excessive QT interval prolongation. To determine whether SPX, GRX, and other recently developed quinolones, gatifloxacin (GAT) and moxifloxacin (MOX), have similar, potentially deleterious, properties we compared these agents in two ways. First, we measured their relative antagonist potency against the rapid component of the delayed rectifier K(+) current (I(Kr)), and second we determined the QT interval prolongation and inducibility of VT and TdP using a well established in vivo rabbit arrhythmia model. All of these agents are I(Kr) antagonists with the following IC(50) values (mean +/- S.E.) for I(Kr) block: SPX, 0.23 +/- 0.07 microM; MOX, 0.75 +/- 0.31 microM; GAT, 26.5 +/- 13.4 microM; and GRX, 27.2 +/- 11.6 microM. All agents also increased the maximum QT interval (mean +/- S.E.) from baseline (241 +/- 10 ms): SPX, 370 +/- 30 ms; MOX, 270 +/- 30 ms; GRX, 280 +/- 25 ms; and GAT, 255 +/- 23 ms. No agents caused TdP during a standard 30-min observation period, but SPX-treated animals developed nonsustained VT (three of six) and TdP (one of six) during an extended 60-min observation period. These findings show that I(Kr) block may be a common feature of many quinolone antibiotics, and that the proarrhythmic consequences vary according to I(Kr) antagonist potency, but are also influenced by additional, unidentified factors.

Sympathetic activation enhances QT prolongation by quinidine

INTRODUCTION

Salt restriction results in endogenous sympathetic activation, and we previously showed that plasma concentrations of quinidine measured after oral drug administration are increased during a low-salt diet. However, it is not known whether, independent of effects on plasma concentration, the extent to which quinidine prolongs the QT interval also is modulated by changes in endogenous sympathetic activity.

METHODS AND RESULTS

In these studies, we evaluated quinidine concentration-QT relations during low-salt (10 mEq/day for 8 days) and high-salt (400 mEq/day for 8 days) diets, with or without beta blockade in normal volunteers. In the absence of beta blockade, the concentration producing a fixed (15%) increase in QTc was significantly lower with salt restriction: 1.2 +/- 0.4 microg/mL (low salt) versus 2.2 +/- 0.4 microg/mL (high salt) (P < 0.01). With beta blockade, this difference was abolished: 1.9 +/- 0.3 microg/mL (low salt + beta blockade) versus 2.1 +/- 0.3 microg/mL (high salt + beta blockade). QT morphologic abnormalities including bifid T waves and U waves were abolished with beta-adrenergic blockade.

CONCLUSION

Sympathetic activation by a low-salt diet not only modulates drug disposition but also increases sensitivity to drug-induced QT prolongation.

Delayed activation and retrograde propagation in cardiac muscle: implication of virtual electrode effects

Point cathodal stimulation of cardiac tissue was shown previously to produce both a dog-bone shaped virtual cathode transverse to the muscle fibers and two longitudinal virtual anodes. We hypothesize that virtual anodes can cause a region of delayed activation, separating two regions of early activation caused by the virtual cathode. Using a high-density electrode array in 42 superfused epicardial slices from 14 canine left ventricles, we observed regions of early and delayed activation and different pathways of retrograde propagation corresponding to the earlier patterns. Retrograde propagation was seen from the transversely located early activation area through areas of delayed activation toward the cathode, and from the early activation area toward the cathode directly. These pathways caused a wide dispersion in the direction of retrograde propagation (2 degrees +/- 31 degrees, n = 179, relative to the fast axis of threshold activation; radial velocity: 0.5 +/- 0.2 m/s, n = 95, in 12 slices from 8 hearts with stimuli of 330 micros, 0.8-30 mA). Delayed activations were observed 0 degrees +/- 6 degrees (n = 32) from the axis in 23 maps (at differing stimulation strengths) recorded in 13 slices from 10 hearts. We conclude that point cathodal stimulation induce delayed activation along the fiber axis and retrograde propagation both along and transverse to the axis.

A common polymorphism associated with antibiotic-induced cardiac arrhythmia

Drug-induced long QT syndrome (LQTS) is a prevalent disorder of uncertain etiology that predisposes to sudden death. KCNE2 encodes MinK-related peptide 1 (MiRP1), a subunit of the cardiac potassium channel I(Kr) that has been associated previously with inherited LQTS. Here, we examine KCNE2 in 98 patients with drug-induced LQTS, identifying three individuals with sporadic mutations and a patient with sulfamethoxazole-associated LQTS who carried a single-nucleotide polymorphism (SNP) found in approximately 1.6% of the general population. While mutant channels showed diminished potassium flux at baseline and wild-type drug sensitivity, channels with the SNP were normal at baseline but inhibited by sulfamethoxazole at therapeutic levels that did not affect wild-type channels. We conclude that allelic variants of MiRP1 contribute to a significant fraction of cases of drug-induced LQTS through multiple mechanisms and that common sequence variations that increase the risk of life-threatening drug reactions can be clinically silent before drug exposure.