NEW YORK (GenomeWeb) — Despite sustained interest among researchers and diagnostics developers in advancing molecular tests to personalize cardiovascular care, integration of such tools into clinical practice is happening slowly where at all.
At the recent annual meeting of the American College of Cardiology, in presentations on using biomarkers for cardiovascular risk prediction, pharmacogenomic algorithms for dosing platelet inhibiting or blood thinning drugs, or the contributions of genetics to a variety of heart and vascular diseases, doctors and researchers sounded a repeated refrain — "we're not there yet."
Unlike oncology, where early genetic and other 'omics discoveries have seen accelerating clinical adoption — in some cases becoming routine elements of patient care and assessment over the last decade — presentation after presentation at the ACC meeting revealed that there has been no similarly easy entry of molecular tools into the kit of the practicing cardiologist.
At the same time, the meeting highlighted that many cardiovascular diagnostic and prognostic markers may be poised to truly prove themselves. But amidst palpable excitement about how molecular tools could improve patient care in the near future, presenters largely called current advances "not quite ready for" prime time.
In a debate on the merits of genotype and platelet function testing to guide treatment with clopidogrel, Sanjay Kaul of Cedars Sinai Medical Center quoted former US Food and Drug Administration official Lawrence Lesko, who bemoaned almost a decade ago in a paper, the "schizoid state of personalized medicine," concluding that the field was simultaneously an "elusive dream," hindered by numerous challenges to its broad adoption, and an "imminent reality."
"I regret to say, but seven years down the line, this schizoid state persists [in cardiology]," Kaul said at the meeting.
Of course, fundamental differences exist between cardiovascular disease and cancer. In cancer, the relationship between genotype, phenotype, and treatment response can often be very clear, whereas in heart disease — apart from distinct subsets of inherited structural defects and cardiomyopathies — phenotype and disease risk are governed by a complex web of rare genetic variants, each contributing relatively minutely.
For example, in a discussion of the potential of genetics to improve atherosclerosis risk stratification and statin dosing at the meeting last month, Sekar Kathiresan, a Massachusetts General Hospital clinical cardiologist and human geneticist, explained that studies over the last several years have found numerous gene regions that are associated with increased risk of coronary artery disease and myocardial infarction, but without clear evidence that considering genotype in clinical practice can actually affect patient outcomes, doctors have little reason to adopt genetic testing.
"Atherosclerosis has a long asymptomatic phase, and there is an effective intervention, namely statins, that might reduce risk during this phase. In the US and in most of the world, current prescribing approaches treat those with highest absolute risk, but as a result we are treating mainly older individuals [essentially] waiting for an advanced phase of the disease."
"This exposes the clinical need for a marker for early identification of high-risk individuals in this space," he explained.
Kathiresan and his colleagues combined some of the first of their discovered risk markers into a multi-gene scoring method and tested it using data from prospective cohorts representing many thousands of patients. The group found that the combined genetic risk score had just as much ability to differentiate higher- and lower-risk patients as clinical factors that cardiologists currently use when they assess patients and decide whether to prescribe statins. The group published this study in the Lancet in 2010.
However, in Kathiresan's view, it's still unclear what the clinical utility of such a score would actually be. "The hypothesis is that early initiation of statin therapy will benefit young but high-risk individuals [and] that those individuals can be identified based on genetics," he said.
"We have shown that this genetic score can identify 20 percent of the population who are at 1.7-fold increased risk. That degree of risk stratification is comparable to other [factors] we currently use in clinical practice, like diabetes status, but we don’t know for sure that it will be useful. … Additional studies will be needed to clarify," he concluded.
Even some of the earliest genetic and molecular factors to enter the cardiology space have struggled to definitively prove their clinical utility.
Stephen Kimmel of the University of Pennsylvania School of Medicine gave an update at the meeting on the publication late last year of several large prospective trials comparing genotype-driven and non genotype-driven dosing of the anticoagulant warfarin.
The FDA updated the label for warfarin in both 2007 and 2010, stating that patients with certain variants of CYP2C9 and VKORC1 genes may require non-standard warfarin doses when they start therapy. But practice guidelines have not changed to recommend genotyping patients ahead of prescribing them the drug, and as such, the majority of doctors have not adopted genotyping strategies into their clinical practice.
In December last year two independent randomized controlled studies appeared simultaneously in the New England Journal of Medicine with different conclusions. One study, conducted by the EU-PACT group and involving 455 patients, compared a cohort prescribed warfarin based on a PGx dosing algorithm with a cohort given standard empirical dosing informed by neither genetic nor clinical factors. This study found that patients in the PGx-guided group stayed longer within therapeutic range compared to those in the control arm (67.4 percent versus 60.3 percent).
The other study, Clarification of Optimal Anticoagulation through Genetics, or COAG, randomized approximately 1,000 patients to receive either a warfarin dose based on both PGx and clinical factors, or dosing only based on a clinical algorithm. Led by University of Pennsylvania's Kimmel, the COAG investigators found that in the first four weeks of treatment, genotype-guided warfarin was no better than clinical factors alone in terms of patients' mean time within therapeutic range.
According to Kimmel, based on the conflicting evidence of these two trials, warfarin has essentially failed to live up to the promise of earlier observational studies, which indicated that PGx algorithms can more accurately predict the dose required to put patients in therapeutic range and keep them there.
While the data has been clear that genotype is predictive of dose, it just hasn't played out in prospective studies that this actually translates to clinical utility. Thus, clinicians cannot currently conclude that the data "supports even incremental benefit of genetics above and beyond clinical factors," he said.
There are also clinicians who would disagree with Kimmel, and believe the patients they see are more like those evaluated in EU-PACT. As such, they believe that genetic testing to dose warfarin would help patients stay longer in therapeutic range than a standard dosing strategy.
A larger trial comparing outcomes after genotype- and non-genotype-driven warfarin dosing conducted by Iverson Genetics, which makes the GenoSTAT Test for identifying mutations in the CYP2C9 and VKORC1 genes, is ongoing with the hope that final data will help resolve these previous studies' stalemate.
Similar issues, meanwhile, were apparent in debates at the ACC meeting about the utility of genomics to inform dosing of warfarin's blood thinning cousin, the antiplatelet drug Plavix.
The FDA issued a boxed warning to Plavix (clopidogrel) in 2010 alerting doctors that patients with certain variations in the CYP2C19 gene have a diminished ability to process the drug compared to those with a normal version of the gene, and these poor metabolizers are at heightened risk for cardiovascular events while being treated with the drug after having acute coronary syndrome or a stent procedure.
However, the label change does not require CYP2C19 typing to guide clopidogrel treatment, nor has the FDA approved a specific companion diagnostic test for the drug. Professional cardiology guidelines have reflected this ambiguity.
Mathew Price, an interventional cardiologist at Scripps Clinic, said in a debate on genetic testing and platelet function testing in patients with acute coronary syndrome that while there is a "tremendous unmet need to individualize antiplatelet therapy," he believes that "current data do not support that we are there yet."
Cedars Sinai's Kaul echoed that in a separate debate on ACS personalized medicine, saying that "results from large randomized trials [of genetic testing] have failed to confirm the promising results of earlier studies."
Marc Sabatine of Brigham and Women's Hospital, Kaul's debate opponent, countered that there has actually been evidence from some trials showing that tailoring treatment with clopidogrel based on platelet reactivity measurements in patients prior to stenting has resulted in significantly lower rates of death and ischemic outcomes than giving a standard dose.
"Clopidogrel continues to be used in a one-size fits all approach, despite the fact that we know there is a high-variability of pharmacologic response and that high on-treatment platelet reactivity is associated with the worst outcomes. Reactivity testing and/or genotyping along with other clinical factors, I think, is logical with the caveat that it should be done in high-risk patients," he concluded.
The FDA's boxed warning in 2010 for clopidogrel came as the drug was close to losing patent protection and newer anti-platelets were entering the market. With growing competition from newer drugs without the PGx variability of Plavix, arguments for gauging CYP2C19 polymorphisms associated with variable response to the drug also lose steam. However, proponents of genetic testing on high-risk cardiac patients in this setting argue that such an intervention can be cost-effective by directing best responders to cheaper, generic clopidogrel.
Sabatine said that it’s a fallacy to think that with the entry of newer, less variable alternatives, clopidogrel is not going to be used in significant numbers of patients. "I'm a big fan of the new agents," he said. "But if you are debating which patient should get clopidogrel versus a third-generation drug, at least one of the factors logical to take into account is genotype."
In his presentation on clopidogrel genetics, Sabatine also implied that the cardiology field may be demonstrating some hypocrisy in requiring genetic or other molecular markers to show greater predictive power than clinical factors that are already welcome tools in diagnosis, prognosis, and prescribing. A statistical measure called the C-statistic, which indicates an overlap in risk between those with a particular marker or feature versus those without, Sabatine explained, is often used to judge a test or model's predictive ability.
"The C-statistic is a very good measure for diagnostic tests, when [you want to determine if] someone has a disease or not [and] you want that to be 90 or 100 percent certainty. But when you talk about predicting the future, it's much harder. Accepted clinical risk factors [like] hypertension, smoking, cholesterol, all things we know have significant risk ratios, these all change the C-statistic in models at relatively trivial levels," he said.
In his presentation about atherosclerosis risk, Kathiresan echoed this as well, comparing the C-statistic for four risk factors — a genetic risk score, LDL, blood pressure, and the biomarker C-reactive protein — from a longitudinal study of several thousand subjects. C-statistics for each were nearly the same.
But in the end, most at the conference agreed that for molecular or genomic tests to find widespread use in cardiology, they will have to prove their clinical utility in prospective studies.
Several presenters at the event hinted at why clinical utility has been much harder to demonstrate for genomics and molecular diagnostics in cardiovascular disease than in the personalized medicine poster child, cancer. In cancer, effective treatments are limited and patient outcomes can depend dramatically on a single test or molecular marker.
Meanwhile, according to physicians presenting at the ACC meeting, cardiologists already practice personalized medicine with or without genetics and risk assessment and diagnosis of cardiovascular disease already requires physicians to weigh hosts of different clinical factors, which they wield like an artist's palette. This makes it much harder to demonstrate what one extra piece of genomic or molecular information can add to that process, and thus what the magnitude of additional benefit to the patient might be.
The challenges the field has seen in demonstrating the utility of genotype-driven warfarin and clopidogrel dosing — even after many years of data showing clear links between genetics, drug metabolism, and associated risks — have highlighted the difficult path forward for emerging molecular markers.