no further comment necessary...

Your estimate of 14,000 stent thrombosis events is misleading.  This would only be true if all 1,000,000 patients had the 2C19 LOF allele. Only a minority of patients have this allele so your estimates wildly overstate the problem.

Over 30% of people of European ancestry are carriers of a loss-of-function allele of CYP2C19 and that percentage is increased further among Asian and African ancestries. The benefit of avoiding stent thrombosis is 14 per 1000, as underestimated in the paper by Holmes in JAMA which exluded studies that reported on stent thromboiss. No matter this  translates to avoiding 14,000 stent thrombotic events per 1 million patients treated.

The *2 allele is at least 25% of the population... so, not such a wild overestimation.

Great concise vlog.  Even a luddite oncologist like me can understand this!  Given the global  predominance of CVD and neoplastic disease in measures of mortality, it would seem we should hold a shared conference on genomic medicine and NCD's.  

In response to Dr. Price's question regarding the pushback to genotyping, I feel that the role of genotyping is limited given that it only explains approx. 12% of the variation in rensonse with clopidogrel.  As much as I would love to incorporate genotype screening as part of routine practice, I still think it is somewhat prematre to use it widely since there are no clear evidence-based recommendations (please correct me if I am wrong!) for dosing alternatives given that some studies have shown that even after a 2400mg load of clopidogrel there may still be an 8% rate of non-responders based on platelet reactivity studies (Bonello et al).  Also, as you very well stated, the FDA will post warnings regarding renal adjustments, however, when the FDA released the boxed warning on clopidogrel non-responders, it did not provide a dosing regimen that would overcome the problem.  The warning noted that other options may be considered.  Prasugrel, the only other antipletelt available at the time, is not a one-size fits all and it has a narrow niche- so this agent certainly does  not "fix the problem". Ticagrelor poses a financial barrier for some patients.  Furthermore, back when the warning was released, GRAVITAS was still in the works and CURRENT -OASIS7 was not of much help. The FDA alerted clinicians of the possibility of a genetic variation that may pose a risk for increased CV risks but it did not provide dosing alternatives. All renal warnings provide a pharmacokinetic-based dosing adjustment for renally impaired patients, which we all trust and rely on.  So I agree with your comments and concern regarding the misleading tone of the meta-analysis in question, because I feel there may be a narrow niche for this genetic testing (mainly in combination with platelet function testing), but I would like to share that most of us are still reluctant to utilize genotyping because of the aforementioned reasons (in my opinion, of course).  I can certainly see the role of genotyping in elective cases where $$ is an issue and your only option for antiplatelet pot-stenting is clopidogrel. In such circumstances, a patient with a faulty allele may not be a candidate for stenting and may elect for medical therapy based on risk v. benefit. However, we must all recognize that genotype (CYP2c19) testing may be misleading because a patient may have wild type CYP2c19 alleles and faulty ABCB1 alleles which may affect the absorption and hence effectiveness of clopidogrel (esp. if it is an obese female...), and yet based on the genetic testing for CYP2c19, you would stent the patient and order clopidogrell and the patient MAY have residual on-treatment platelet reactivity post-stent which translates into worse outcomes (all of which would be missed with genotyping).  Just my thoughts... Thank you for a very informative spec. edition Topolog!

In the meta-analysis, the actual rate of LOF alleles was 9.5% (see figure 2, first column on the left).  Dr. Topol, et al. overestimate the effect of this 11-fold.  Even if your's and Dr. Topol's estimates are correct, the estimate is off by a factor 3-4.  It would only be 14,000 stent thrombosis events per million LOF patients, not all-comers.

Regardless, the more comprehensive outcome endpoint there was no benefit suggesting that any increase in stent thrombosis was offset by decrease in other events. This seems implausible. I believe the endpoint which is more comprehensive and clinically important (and not to mention with the larger N).

Aside from all of this, the genotype is only one of many factors which affects rate of stent thrombosis. It is still a long way to showing clinical utility given the importance of technical factors (such as stent apposition) and patient factors (drug adherence, interactions, etc.). Cost effectiveness will be an even higher burden given the very low rates of events and the fact that even with recognition of genotype, the excess stent thrombosis might not be completely avoidable with current therapeutic options (see GRAVITAS).

The issues dicussed in this forum are far and above pharmaco-genomics of clopidogrel.

The issue is the future of Genomic-Personalized-Privatized Medicine(GPPM)

Today ,Medicine is in crisis:Economic and Ethic.Simply said:We cannot afford

this beautifull monster.The human society cannot comprehend the importance of

personalized profile of every disease,biochemical pathway,proteinomic,receptor

etc.All tailored to each person individualy.

Today,people look at Medicine as an awesome good looking monster,that it is beyond their reach:intelectualy and financialy.

In order to make medicine within the reach of our patients.We need :Affordable

Medicine at the level of molecular biology-molecular medicine.

This is so true about Clopidogrel,a medicine that we are so proud of.Still we have so many doubts about its value,with all its side effects.Does it,rely fulfill its mission?

That is,unfortunately so true concerning many other drugs.

How can we corroborate clopidogrel response with bedside platelet function testing? Especially when we know that the pharmacological response to clopidogrel is heterogeneous. I agree with Dr.Topol that only those who need to undergo stenting may be tested.

The author’s conclusion that “overall there was no significant association of genotype with cardiovascular events” does not appear to be supported by either the data or their own analyses.

 

The overall summary point estimate is an 18% increased risk of CVD events among 2C19 carriers (1.18, 95% CI: 1.09-1.28) using a fixed effects model and a 34% increase in events (1.34, 95%CI: 1.15-1.56) using a random effects analysis.  The authors disregard the summary estimate in favor of the 4 largest studies that show no effect because of evidence for publication bias (the potential for preferential publication of small studies with large effects by chance).

 

However, in further analyses, 2C19 carrier-status was associated with an increase in each of the individual endpoints:  fatal of non-fatal MI: 1.37, 1.13-1.65; non-fatal MI: 11.48, 1.05-2.07; stent thrombosis: 1.75, 1.50-2.03.  Although greater among the smaller studies, each of these point estimates was significant in both studies with fewer than 100 events, and in studies with greater than 100 events.

 

The appropriate interpretation of these data, overall, therefore is that 2C19 carrier status appears to be associated with an increased risk of CVD events among persons being treated with clopidogrel, but the magnitude of this effect may be over-estimated due to the effect of publication bias. 

 

In addition, the authors show that in the 4 placebo-controlled RCTs 2C19 carriers derived only ½ the RRR of events during treatment with clopidogrel than did non-carriers (RRR: 13% v. 22%).  They dismissed this “effect modification” because it was not numerically stable (p=0.37).  However, detecting effect modification is limited by low power.  The absence of evidence for effect modification (or interaction) is NOT evidence AGAINST effect modification.

 

This last point is very important because methods for evaluating genetic interaction and effect modification are likely to become increasingly important for personalized medicine.   In this regard, I would like to refer theHeart.org readers to a brilliant new paper in PLOS ONE on genetic effect modification:

 

A Common KIF6 Polymorphism Increases Vulnerability to Low-Density Lipoprotein Cholesterol j.mp/vse1Dq

 

The authors show that the KIF6 SNP is not associated with the risk of CVD and it has no effect on the LDL lowering effect of statins. Despite this, they show that KIF6 carriers do appear to experience a greater reduction in events during treatment with a statin than non-carriers.  The authors employ an original and clever use of meta-regression and a novel interpretation of genetic effect modification to show that this is possible because KIF6 increases vulnerability to LDL, and as a result KIF6 carriers experience a greater clinical benefit from lowering LDL than do non-carriers.  The study shows that our genes may influence our vulnerability to LDL and other CVD risk factors, and therefore it introduces an entirely new way of thinking about what pharmacogenomics means and how we can potentially use SNPs to help choose treatments for our patients. This study shows the kind of original and important thinking that we need to move the field of personalized medicine forward.  It stands in direct contrast to the current study, which too easily dismisses a potentially important pharmacologic effect of 2C19.

 

Happy New Year to all.

 

 

Spot-on analysis, Truman.  Well done.

I would only add that because of the heterogeniety in the types of trials included, the types of outcomes included, and the indications for treatments included, a random effects model is probably more appropriate.  This presents a real problem for the authors because the addition of a random effects term mu has the effect of weighting each study more similarly than a fixed effects model. As a result, smaller studies have more weight in a random effects model than in a fixed effects model.  The authors however are concerned about small study or publication bias.  In my opinion, they make the mistake of resolving this methodologic dilemma by presenting the fixed effects result (with its smaller point estimate) and then simply concluding that the association is probably not real after accounting for the effect of the effect publication bias. 

Lets keep the discussion elevated by focusing on deconstructing the data and methods, instead of attempting to deconstruct each others motives.

Thanks for the KIF6 article link, really innovative stuff!  If we can predict clinical response based on a patient's genetically determined vulnerability to a modifiable risk factor that could have a really profound impact on medicine. 

 Genotyping and their relationship with clinical outcomes: Stents: CYP2C19: Factor V Liden.   Help me understand. I have had 3 open heart surgeries, aortic valve replacement, so on coumadin, plavix, and aspirn.  Still got DVT in legs (where next?) 

Also, mother has had more than one stent, was a smoker, COPD, on Plavix since 2009.  Will be checking to see if she has Factor V Liden and (CYP2C19 ?) 

And if there had been Genotesting, would some of these issues been avoided?  Just curious about this subject. 

 

Marijo Wheeler Giardina

831-905-0414

Ryan 

This would be nice to add to our media piece directed to physicians.