UPMC Physician Resources
ORLive member exclusive presentation. Please sign in below to view the video.
Membership is free and will provide you with access to this video and the option to receive information about new content on ORLive.com
DOM Medical Grand Rounds: The Bench to Bedside: The Future of Pharmacogenomics and Precision Therapeutics
In this presentation, Drs. Dennis McNamara and Philip Empey each present from the department of medicine medical grand rounds. Dr. Mcnamara discusses heart failure in African Americans and provides an overview of the unique epidemiology compared to white counterparts and how medical therapy and pharmacogenetics can be used to improve the distinction. Dr. Philip Empey explains some of the new initiatives going on at the University of Pittsburgh and lays a framework for how more research can be conducted in the future.
Upon completion of this activity, participants should be able to:
- Improve the ability to diagnose the etiology of heart failure in African Americans.
- Improve outcomes by increased utilization of proven heart failure therapies
- Improve the potential of genomics to target therapy in the future.
- Describe why pharmacogenomics is a leading use case for precision medicine.
- Integrate genetic data with clinical variables to guide antiplatelet medication prescribing following percutaneous coronary intervention.
- Improve the ability to diagnose the etiology of heart failure in African Americans.
- Improve outcomes by increased utilization of proven heart failure therapies
- Recognize the potential of genomics to target therapy in the future
1. Yancy CW. J Card Fail. 2000;6:186.
2. Hare JM. N Engl J Med. 2004;351:2112-2114
3. Carson P, et al. J Card Fail. 1999;5178-187.
4. Rosskopf et al. Hypertension 2000 936) 33-41.
5. McNamara DM, et al. JACC Heart Fail. 2014; 2(6):551-557.
6. Ware JS et al. N Engl J Med 2016;374:233-241.
7. Meyer UA. Nat Rev Genet. 2004;5(9):669-676.
8. Empey et al. Crit Care Med. 38(6):s106-16.
9. Shudliner et al. JAMA 2009;302(8):849-857.
10. Scott et al. CP&T. 2011;90(2):328-32.
11. Kim et al. CP&T. 2008;84(2):236-242.
12. Mega et al. NEJM. 2009;360(4):354-362.
13. Simon et al. NEJM. 2009; 360(4): 363-75.
Dr. Empey has reported no relevant relationships with
proprietary entities producing health care goods or services.
Dr. Mcnamara has financial interests with the following proprietary entity or entities producing health care goods or services as indicated below:
- Grant/Research Support: Arbor Pharmaceutical Co. Provide Drug for GRAH2
- Consulatant: Sanofi Pharmaceutical, GE Healthcare
All presenters disclosure of relevant financial relationships with any proprietary entity producing, marketing, re-selling, or distributing health care goods or services, used on, or consumed by, patients is listed above. No other planners, members of the planning committee, speakers, presenters, authors, content reviewers and/or anyone else in a position to control the content of this education activity have relevant financial relationships to disclose.
The University of Pittsburgh School of Medicine is accredited by the Accreditation Council for Continuing Medical Education (ACCME) to provide continuing medical education for physicians.
The University of Pittsburgh School of Medicine designates this enduring material for a maximum of 1.0 AMA PRA Category 1 Credits™. Each physician should only claim credit commensurate with the extent of their participation in the activity. Other health care professionals are awarded (0.1) continuing education units (CEU) which are equivalent to 1.0 contact hour.
For your credit transcript, please access our website 4 weeks post-completion at http://ccehs.upmc.edu and follow the link to the Credit Transcript page. If you do not provide the last 5 digits of your SSN on the next page you will not be able to access a CME credit transcript. Providing your SSN is voluntary.
ABIM MOC Part 2 Credit
Successful completion of this CME activity, which includes participation in the evaluation component, enables the participant to earn up to 1.0 MOC points in the American Board of Internal Medicine's (ABIM) Maintenance of Certification (MOC) program. Participants will earn MOC points equivalent to the amount of CME credits claimed for the activity. It is the CME activity provider's responsibility to submit participant completion information to ACCME for the purpose of granting ABIM MOC credit.
To receive your ABIM MOC Part 2 credit, you will need to complete the post-test and evaluation process and provided your date of birth and ABIM number, along with the other required fields.
Release Date: 2/2/2017 | Last Modified On: 2/2/2017 | Expires: 2/3/2020
Heart failure in African Americans is a very important clinical problem, very common unfortunately, has a unique epidemiology. It's a bit distinct from the heart failure we see in the white counterpart. There is a lower incidence despite risk factors of coronary disease and there is a higher incidence, much higher incidence of hypertension. Unfortunately and the reason it's so important is there is a worse prognosis, higher rates of hospitalization, higher rates for the same ejection fraction of sudden death and increased mortality even on multivariant analysis both for African American men and women. And how can we use medical therapy and pharmacogenetics to really improve this unfortunate distinction? There is clearly a distinct response to medical therapy to the point where we now have heart failure medications that are actually approved separately for self-designated African Americans.
Now I would argue, and we'll show there are investigations to show that the heart failure phenotype, the distinctions in the heart failure phenotype particularly when it comes to hypertension are genetically based, that some of those same differences affect clinical outcomes and potentially are targets for therapeutic intervention.
Now this next slide shows data from a number of multicenter trials about the incidence of -patients with systolic heart failure in our trials. We have 1 VHeFT, II are some of the oldest VA trials, SOLVD was a good ace inhibitor trial, Carvedilol, BEST and MERIT are the trials that really made beta blockers standard therapy. And then if you look in the subjects in those trials those, if you look at the white subjects which are primarily in purple, they had primarily coronary disease as etiology of their heart failure; while if you look at the black subjects in red, the bottom part of the slide, they predominant had hypertension. It somewhat mirrors what we see in the clinic, it's a little bit of a different phenotype. That has led to studies of genomics going back 2 decades now to try to understand that difference.
The other thing we know in terms of heart failure is left ventricular hypertrophy seems to be more pronounced in our black subjects with heart failure than in white counterparts. This is from a group in the Dallas Heart Study of self-designated African Americans not treated, so at time of presentation, that looked at the correlation of LV mass with blood pressure presentation, of course it goes up. But look at the distinction by the presence or absence of a genotype of Corin which is really only seen in black cohorts and not seen in whites. Corin is the converting enzyme for BNP. So it's felt natriuretic peptides which are cardioprotective, so a deficiency in production more common in African Americans leads to greater LV hypertrophy for any given blood pressure.
Now we have been interested in the Heart Failure Center University of Pittsburgh on genomics and particularly on myocardial recovery and nonischemic cardiomyopathy for about the last 20 years. And we've done a number of multicenter trials.. I'm going to show some data from IMAC2, Intervention of Myocarditis and Acute Cardiomyopathy, which is a study of all recent onset nonischemic myopathy both men, women and women with peripartum disease at 16 centers completed about 5 years ago. And I'll also show some data from IPAC, that's a more recent study at 30 centers specifically in women with peripartum cardiomyopathy.
Now this is from IMAC2, so this is the ejection fraction at baseline when they enter. All patients with new onset nonischemic myopathy and 6 months by race. The 80 black subjects, 293 white subjects, roughly similar ejection fractions at presentation, both about 23, 24%, but at followup there was less recovery in the African American subset. And this is despite similar therapy.
If you look at outcomes again this is a survival free from heart failure hospitalization. There is also a distinct difference based on race with much poorer event free survival in black patients compared to whites. And again I don't think that this is a treatment difference. Treatment of both groups was similar in terms of ace inhibitors and beta blockers. This is at entry but it's actually even closer at 6 months, and this timeline is their time to ICD implantation, so just to give a sense of how people were managed and it looks virtually identical in black and white subjects across this 16 IMAC centers. So therapies were quite similar but outcomes were distinct based on race.
This is from our peripartum study, so 100 women across 30 centers and peripartum cardiomyopathy is an unfortunately common complication, or rare but unfortunate complication of pregnancy, unfortunately a common cause of maternity mortality. The women presented with the black subjects on the right side of the screen, about 30% of the cohort was black; and the nonblack, mostly white patients on the left side. The black patients came in with a lower mean ejection fraction and just never caught up. Again similar treatment is the best we can say just came in a little bit more severe LV dysfunction and just don't catch up.
This list which is growing every decade is the therapies and this is part of the problem we have. These are the therapies which improve survival in heart failure with reduced ejection fraction. We have, we have 7 different combinations with 9 different therapies growing every day, but no patient in the world is going to take all of these. Which ones work for which patients? The oldest therapy and the one I’m going to talk more in detail about is Hydralazine in combination with Isosorbide dinitrate, the oldest and first therapy to improve survival in patients with systolic heart failure.
Now this has a relatively unique mechanism compared to most of the other drugs which are working more directly on renin angiotensin system. And we believe it acts primarily through NO, so Isosorbide dinitrate is a organic nitrate that stimulates NO production. There is a physiologic pathway which is cardioprotective through cyclic GNP as well as through posttranslational modification that is certainly beneficial in heart failure. There is also a pathologic pathway where it can combine with the superoxide ion to form peroxynitrite which has a detrimental effect. Now while Isosorbide dinitrate stimulates production of NO we believe Hydralazine is an inhibitor of the oxidase production and we feel drives it through a more protective cardioprotective pathway. As I said this was the first therapy shown to improve survival in systolic heart failure.
The classic VHeFT 1 is a VA study and when that study showed, demonstrated improvement in survival in the top of the screen what we didn't know at the time was there was a much different response by race. So in VHeFT 1 the nitrates and Hydralazine combination in orange improved survival, so a lower curve, but overall it was a positive trial but when you broke down the substance analysis it was only effective in the black subset and there was very little benefit in whites.
Subsequently that was followed by VHeFT 2 which was one of the first studies to say aha ace inhibitors are the real savior in this circumstance and standard of care and have been that way ever since. But when you looked back in VHeFT 2 and did the same breakdown the treatment group of Enalapril was superior only in whites. And the only randomized study ever done, VHeFT 2, of nitrates and Hydralazine versus ace inhibitors in black subsets showed no difference.
This led to the rationale for the A-HeFT trial which is now over 10 years old which looked specifically in a cohort with systolic heart failure which were self-designated at African Americans, and used it on top of the all of the therapy and was stopped early. It's a small trial by heart failure standards, it's only 1,000 people, 500 in each treatment group, 160 centers, so very difficult to do. But was stopped early because of a huge survival benefit of 43% reduction in mortality of patients on therapy. Now this is over 10 years. Now what percentage of African Americans are currently on this combination therapy shown to reduce mortality by 40%, nationwide about 20%. 80% of African Americans with systolic heart failure are not on this therapy. Think about if we had the same thing with ace inhibitors or beta blockers? We'd be like all up in arms. Why? I don’t think it's just doctors not being aware, right now this is additive therapy and it's hard to get your patients to take their 6 and 7th medication.
Now we were the Genetic Core Lab for A-HeFT. And as part of the theme you know all therapies do not work for all people we tried to look at what the genetic differences were and if we could tease out why this therapy was more effective in African Americans and if we could tease out a genomic signature rather than a racial construct for determining therapy.
Now this is a schematic of the renin angiotensin pathway where we begin with adrenalin and drive forward through AII as well as aldosterone production. This where most of our therapies, and I would argue that Hydralazine and nitrates also through AI works on several of these points as well. All our therapies, older therapies which improve survival act on this point. And there are significant polymorphisms, genetic variants which affect the functionality of each of these points and I would argue also affect the impact of our medical therapies.
In GRAHF, the genetic substudy, we looked at a panel of SNPs which have been either been shown to affect heart failure in other cohorts and the majority of these are differentially prevalent in black versus white cohorts and studied extensively in hypertension, we are just trying to apply that to heart failure.
And I want to focus for probably the rest of the talk on GNB3. GNB3 is a gene, it's a G-protein subset which is an important subunit which is important in Alpha adrenergic signaling. There is a common polymorphism studied mostly in hypertension where the T allele is associated with increased signaling and low renin hypertension. And it's much higher in prevalence in black cohorts than in whites.
How does it work? From the schematic it's actually just a marker, it's 100% the solid mutation but it's in 100% dysequilibrium with a splicing variant, it's really the T haplotype. And the splicing variant in Exon 9 results in increased signalling.
It's very different in black and white cohorts. In black the cohort from A-HeFT and from GRAHF about 50% of the subjects are homozygous for the T haplotype where it's only present in about 15% of whites.
We have seen in other cohorts that this same T allele, and this is in a predominantly white cohort, this is from GRACE presented at the AHA, currently being submitted for publication, submitted last year, from our clinic where in a predominantly white cohort the TT genotype in blacks - excuse me in the black line here in a mixed cohort was associated with poor survival.
We had seen the same thing previously, and this is the TT genotype now in red, in the IMAC II cohort. Now less significant, less events but the same prevalence.
What did we see in A-HeFT? Did it affect the effectiveness of therapy? Well it did to a great degree. The primary end point for A-HeFT was a composite score that involved both survival, hospitalization and improved quality of life. And that led to a composite score, if your composite score was high that's a good thing, if it's low that means it's bad. If you look at the improvement on therapy in red in the group with the TT genotype it was very strongly positive, very effective. And if you looked at the other half of the cohort without the TT genotype we saw very little effective therapy.
Now this is just event free survival. Similar if you looked at event free survival by treatment, within the TT subset you saw significant impactive therapy on the left side of the screen, but not in subjects without the genotype. Now this is a small cohort which kind of just it begs some type of validation study and we are currently involved in that validation.
We are coordinating that which is GRAHF2 which is collecting another cohort of up to 500 subjects of self designated African Americans with systolic heart failure who we will treat with a fixed dose combination of Hydralazine and nitrates and try to generate the same A-HeFT composite score to try to see if we can validate that previous study. We do it at the 20 centers throughout our A-HeFT, excuse me our GRAHF network, but that's being coordinated through our Heart Failure Center at University of Pittsburgh through a grant from the NIH.
We hope to also look for other loci which may explain the differences in drug effect. This looks at admixture analysis and just points out the fact that most - that race is kind of a social construct and that most self-designated African Americans have a combination of African genomic ancestry and European genomic ancestry. From A-HeFT 2/3 of the cohort were between 5% and 20% European, and you can use that admixture to actually do genomic screening with a large enough cohort.
And putting multiple loci together even in A-HeFT1, GNB3, NOS3, Aldosterone Synthase, Corin is probably the way to go in the future. This just shows an example of that that you can really try to by using more than one loci you can really try to tailor your therapy to an individual.
Now in the last couple of minutes I just want to show the importance of this same genotype in myocardial recovery and use our peripartum cohort to do that. As I said a major cause of maternity mortality, phenotype identical to other forms of nonischemic myopathy but specifically occurring in young women of childbearing age around the time of delivery. It's increasing in occurrence I think because of recognition. It involves several thousand women each year in the U.S. and it's an important unfortunately problem.
Now there is a marked difference, race is a risk factor. It's more common among black women than among whites. In our cohort in Pittsburgh and nationally we have 10% of an African American population in the city, about 30% of my patients are African American with the disorder. It's also much more prevalent in Africa and in Haiti and I do think that African genomic risk factor, as is with other nonischemic myopathies, is a risk factor - is a problem, is a risk factor.
So for IPAC we looked at these 100 women across 30 centers. And this is the same data I showed you before but in a different way, and again black women in yellow came in with a lower ejection fraction and just don't recover as much at 6 and 12 months with a mean that's almost 10 points lower than their white counterparts.
I think the same genetics are likely at place here and this from a recent publication in Circulation Heart Failure from our group shows now the recovery baseline 6 and 12 months by genotype. So the GNB3 TT genotype in white, other genotypes in the purple and you see that they begin with a little difference but by 12 months it's more than 10 EF units. This is by genotype, not by race. It was statistically significant in both racial subsets and was actually more prominent in the black cohort. As you see there is a difference of almost 15 EF units by 12 months, much more powerful in this subset. But in the white cohort also significant, also significant.
There are other genetics obviously involved here and we recently published this earlier this year in the New England Journal for the IMAC-2 and IPAC investigators' findings of the prevalence of mutations of titin or other genes which cause cardiomyopathy in familiar cardiomyopathy. What is their prevalence in the sporadic disease? This was a collaboration with Cricket and John Sybin out of Boston who had previously reported that in sporadic validated cardiomyopathy, not familial, that you could find titin mutations in 12 to 13%. So we looked in that not only in the IMAC and IPAC cohort but in 4 other smaller cohorts, two from Pennsylvania, one from Germany, one from Japan and what we saw particularly in the IMAC and IPAC cohorts is that of these sporadic cases, these PPCM cases, 12 or 13% had titin mutations, and 18% had some mutations. These are truncations, changing the protein despite the fact that they didn't have a family history. Indeed, 10% of the women in our PPCM cohort had a family history of dilated cardiomyopathy but it was a different 10% it turned out to be. If you add that you'd get almost 28% with some genetic etiology. Now again this is titin, it's a large protein and the mutations both of DCM and peripartum were primarily in the A-band and were about the same prevalence, both about 12 to 13% in both.
What's its impact in recovery? Now this is adapted from that paper. The women with the titin mutations, and this is specifically IPAC in yellow, did seem to recover less, but this doesn't tell the whole story. And indeed if you looked at the white cohort of the IPAC the titin mutation didn't seem to affect it, now 6 and 12 months. How about the black cohort? Well that's where you are seeing the impact. But as I said I think race in this regard is just a marker for genetic background, so let's go and all that data was in the paper but that racial difference made us downplay I think appropriately that we didn't understand exactly how titin was affecting recovery.
But let's do it differently, don’t do it by race, do it by GNB3. And this is very small but if you look at without the TT genotype this includes now some blacks and some whites no effect of titin. Now co-inherit this genotype and you see a much more profound effect. The numbers are very small, we have in the planning stages awaiting funding a much larger cohort and hopefully we'll tease this out. But should we be targeting differently?
So finally genomics and heart failure in African Americans, there is a distinct phenotype, more hypertension, more hypertrophy. I think all of these differences are predominantly genetically based. The same genomic template can help us predict outcomes and I would hope would lead particularly with GNB3 TT genotype will be a target for future intervention.
So thank you for the invitation Dr. Gladwin, and thank you for the opportunity to speak with you today. As I said my passion is in pharmacogenomics and all too often we are selecting medications for patients where we are essentially doing trial and error, maybe with some limited clinical guidance. And the data at this point suggests that we can do better, that only about a half of our patients are responding to therapies in several disease states and I think the excitement of what we have to offer with newer technologies really paints a nice picture for what might happen with precision medicine. So today I'll explain our initiatives here at the University of Pittsburgh and UPMC and overlay a framework for how we can conduct more research in the future.
So why focus on medications? These are the broad areas of interest in terms of deploying genomic medicine, you can see oncology of course, peritoneal screening, complex diseases. Some of the leaders in this institution and internationally really are actually in this room here today. But for pharmacogenomics I think it's a leading use case for the reasons depicted. For most medications we understand how they work, during the drug approval process we understand their common pathways and metabolism. We can generally easily measure drug response or in essence measure drug concentrations in the serum. These are mostly germline polymorphisms so in many cases they do not change, which means the information has lifelong utility. The variants are generally common, so these aren't 1 in a million, they may be 1 in 3 in some cases, and 1 in 10. Testing is generally feasible and cost is decreasing rapidly as many of you are I'm sure aware, and then there are much fewer ethical concerns relative to disease prediction genes. So in reality deploying these requires much less of an infrastructure surrounding understanding what to do with variants of unknown significance.
So this picture illustrates perhaps what we do when we try and treat our patients. In this picture we can perhaps see men and women, it may be fairly difficult to see any difference in populations. In reality our patients are pretty different and really what we want to be able to do is take those mixed populations and before we give the first dose of the first medication aggregate them into these a priority bins. Essentially now before we give that first dose were our patients likely to have an adverse drug event or have therapeutic failure or hopefully get them all up into that top left blue box where we see the majority of our efficacy with minimal toxicity.
Why is this important? Well as we heard in the introduction last year we spent approximately $320 billion in the United States just on drug therapy alone, just on prescription drugs. This is a report from the FDA that suggests that we could do better in terms of getting better drug responses. The majority of our patients actually medications don't work for them. As high as 50% in some disease situations and I would argue that if you want to sort of be on this graph someplace you'd much rather be in the green folks rather than the red folks. Essentially we had a lot of unexplained variability in drug response and perhaps we need to find new tools to try to understand what's driving that variability.
This is where I like to interject pharmacogenomics as a possible solution. This is a very simple graph of Cytochrome P450 2D6 and how genotype correlates all the way to phenotype. Honestly this graph taken from Nature Reviews Genetics was about 20 to 25 years of research. So I'm not going to pretend that we can do this for every medication but it does show the nice association with a particular genotype, so in this case depicted up here in the top, so up here along with the predicted phenotype everywhere from having very poor metabolism with the medication to an expected higher metabolism rate, the frequency in populations in this case for Caucasians and then a mapping to the pharmacokinetics. So in this case we see metabolite ratio or the ratio of parent molecule to metabolite, and you can see in this situation some patients have very high ratios and we see patients way on the other side with low, these fringe phenotypes.
And just moving on we also see that rather than having this bell distribution you might expect we tend to have a bimodal one, suggesting that there were phenotypes within our main populations. And it would argue for perhaps everyone shouldn't get 100 mg of Nortriptyline as depicted on the graph here. Some may require 5 fold lower or perhaps 5 fold lower medications dosing from the very beginning. This is a great example, this is a tricyclic antidepressant. Typically we start these medications and we wait 4 to 6 weeks for a response. Wouldn't it be nice to know if your patient required 50- mg or 500 before you gave that first dose? So a very simple example but I think illustrative and we can pull many other ones from the literature.
So what does that literature look like? This is what sort of keeps folks in my field employed. You can see the number of citations that have come out, the exponential rise in the last - just in the last decade. You remember the completion of the Human Genome Project in 2003 officially completed here and then this was when the NIH pumped a whole bunch of money into the pharmacogenomics research network I would argue with great results in terms of understanding our pharmacogenomics basis of drug response.
So these are the results. Right now in the leading repository of pharmacogenomics information is about 2700 associations between genetic variation and response. Now not all of them are at the highest evidence grade, we have about 34 of them that are 1A evidence, meaning that either they've already been implemented in a large academic health center or we have prospective RCT data saying that this is something that is good for our patients. We have 23 international consensus guidance documents on guidelines of how to actually put this into practice. And then even in other groups, this was the Dutch Pharmacogenomics Working Group, there is large numbers of drugs that perhaps go down to in other areas of the world where other drugs are approved some great data.
Now interestingly this isn't drugs that are used uncommonly. You know these are - you probably recognize them. These are drugs that are in the top 10 of our prescribed medications still to this day. Clopidogrel just 5 years ago was number 3 drug prescribed in the United States, still very commonly used in our cath lab populations here at UPMC and elsewhere. Warfarin still top 10 for adverse drug events, although we do have newer agents on the market it's still the predominant anticoagulant used both at UPMC and across the country. And you can see, you know you can choose your favorite drug here, the list goes on and on, I just pulled some off the list.
The other thing that's exciting to me is you can already start to see duplication. Antidepressants are 2C19, so is Clopidogrel. That means if we test for Clopidogrel we may learn something about other drugs and you can start seeing how this pleiotropy or understanding how one characteristic may affect others could be used to our benefit in treating patients. I mean you could argue that's all sort of in the research sphere and I would argue to the contrary if you look at our FDA approved drug product labeling there's 141 drugs on the market right now that have some sort of statement about pharmacogenomics in them. Now some of them may be very buried down in the pharmacology sections or in the adverse event sections but some of them are right at the top, boxed warnings that are the basis for either a legal requirement to do something for our patients or certainly prescribing guidances for patients.
The other thing that's interesting here is you might assume it was all oncology, and that's really not the case. So oncology is down here in the gray where about 30% of the drugs meet that category, but I'd actually argue that the lion's share of the evidence is outside of oncology in other areas of complex diseases. So you should be thinking internal medicine practices, in private practices out in the community and any of these disciplines that there may be something that we should know about the medications we are prescribing for our patients.
So a simple example. Clopidogrel probably most of you are very well aware of and very exposed to on a daily basis, you can sort of pick your poison here. The left hand side is a schematic of the biology, or if you are chemistry inclined the pathways of elimination are on the right hand side. Interesting, Clopidogrel when we take it is a prodrug, it has little effect, no effect on its own, it needs to be activated through a series of sequential steps to reach an active metabolite. These steps are governed by Cytochrome P450s, the dominant one here is noted, CYP2C19, through a 2 stage activation produces an active ingredient that goes on to bind to our P2RY12 receptor and inhibit platelet aggregation. So most importantly you can think of patients that just came in with a new cardiac event, perhaps got a new drug alluding stent in our cath lab here at Presby, and then we send them off in the world with a medication that's designed to protect that investment and protect their life in the future. I would argue that we really would like to think that a medication is going to work for them.
So where is the evidence for genetics? This is a simple study by Ellen Shuldiner's group, actually a little bit further out west, this is further out east in our own state and into Maryland, this is looking at an Amish population and trying to understand the association between an event, in this case a major adverse cardiac event in a large GWAS study. So those that haven't seen this before this is called the Manhattan Plot because you can see it looks like a skyline at the bottom and these are all the variants being analyzed. So lined up by chromosome and the level of significance with the outcome, in this case a major adverse cardiac event. And this is what we love to see, ti rarely happens in genetics, but we have this single large association at a single locus where we see a clear association with an event. This is a retrospective study that was in JAMA a few years back but really paved the way to say well maybe there is something going on here that we need to evaluate prospectively.
So in those studies we've learned that the frequencies of these variants, the folks with a loss of function or a poor intermediate drug metabolizing enzyme, it's actually fairly common in Caucasians having one of these alleles and most evidenced here by *2 but there is *2 all the way to *8 that could be present. It's about 30% of patients. And then the other end of the spectrum there is a proportion of patients that have an ultrarapid metabolizer predicted phenotype holding at least one copy of the *17 allele.
How does this relate to pharmacology? This is our PK and PD, Pharmacokinetic and Pharmacodynamic associations. In the red there is our poor metabolizer, so this is a patient as you expect, give him a dose of medication, they have a much higher parent drug level versus a patient that is normal. And we look over at the response rate, for those that can't see this is inhibition of platelet aggregation. We expect it to be high for this drug to be working. I don't know about you but if I had a brand new stent and had a new event I would prefer to be on this black line at the top, meaning my drug has a much better possibility for functioning than being down here with 20% inhibition. These are prospective studies doing pharmacokinetic analyses that were published about 5 years ago showing the mechanistic association is able to be demonstrated.
So let's just pull through to outcomes. Again these are retrospective data right now but essentially we know that having one copy of an allele increases your risk of one of these major adverse cardiac events and a 3-fold increased risk of having stent thrombosis after, after your PCI for example. The dose effect, meaning if you are homozygote for a loss of function allele you have a 4-fold increased risk of having one of these major adverse cardiac events and seeing this retrospective data the FDA took an unprecedented step of actually putting a boxed warning right on the label of Clopidogrel at the height of its use when it was number 3 drug in the market. And this is what that says right now. It says diminished effectiveness for poor metabolizers. Poor metabolizers treated with Plavix, the brand name, at recommended doses have higher event rates. Tests are available. Consider alternative therapy in patients who are known to be poor metabolizers. What does it not say? It doesn’t say we should test our patients. The FDA, appropriately, leaves the practice of medicine up to the folks in this room. So it says if you do have data on hand you should use it but when - it is up to you to determine whether it meets clinical utility, perhaps value, perhaps being able to accomplish it in populations to know when to actually integrate it in your populations.
Prospectively this was the largest data published to date, this is from a single institution. This is from the University of Florida published last year and this is major adverse cardiac events at 30 days, so an early outcome after patients had had PCI. The majority of these patients did have acute coronary syndromes and we break them up by whether they had a loss of function allele or not you can see how they segregate. And if you look at outcomes at the bottom you see patients that are normal without a loss of function allele have about a 4% event rate, which is what you expect to see in a population, and what is being replicated in large retrospective studies.
The interesting thing is over here in the left though. So patients that were - had a loss of function variant, the guideline in the institution suggested that the prescriber change therapy, but it was not formally protocolized, it was not forced. They left the prescriber to make the decision on their own. And although this is not a randomized controlled trial it's important to know there may be selection bias here. It does have some interesting outcomes. In a few numbers of patients, only 300 and only 5 outcomes at this point we see a much higher event rate in patients that were not changed on therapies, essentially not increased to the higher potency agent that would not have this expected result. Now one expects that big zero to persist, but it does have some interesting results suggesting that perhaps we can reduce the risk, perhaps not the zero, maybe over to 4%, but certainly down from an expected level of 12 or 13% for an event rate in folks that we are essentially giving a placebo to.
So what happened when this data was published and what's happening across the country? We are certainly a leading academic health center here at the University of Pittsburgh Medical Center, but what's everybody else doing regarding these results? How much has it made it into clinical practice? Well these are the published data. I would argue that Vanderbilt probably was the earliest adopter, they are at about 10,000 in terms of deploying this particular variant currently but you can see other institutions. So this is Florida as I mentioned before, Chicago, other Chicago, Maryland and the other institution there is St. Jude's. In the center is actually really interesting because this is a community pharmacy intercepting your prescriptions when they leave and a patient walks into the pharmacy with them and saying whoa, what, have you been tested yet? You need to be. And if it comes back a certain way we will contact your provider and try to intercept it. And perhaps even most interestingly, that top one up here, that's a pharmacy benefit management company working for an insurance payer. And I would argue if they believe there is value in it, even though they are much further downstream and much further beyond some of the early outcomes, someone is starting to think that there is value potentially associated with this. So this is where we stand in published trials. There is about another 12 institutions out there that are currently implemented either in clinical or research modes and we'll talk more about those there as we talk about what we are doing here at the university.
So what are we doing? This is our press releases from last December when we launched the Pharmacogenomics Program here at UPMC. And this is the design of it. So it's a very simple you know patient/clinician diad. We have clinical data we always typically use for selecting drug therapy, we add in genomics, we feedback recommendations and then really importantly we collect the phenotypic outcome related information in order to create this learning healthcare system we keep all thinking about.
Here is perhaps the maybe too tortured acronym but it does study, and the initiative is called PreCISE-Rx and the goal is to improve the safety and the effectiveness of medications. And overall not only do we have clinical implementation but also a research overlay and several education programs here that are helping clinicians make the best decisions for patients.
What does it look like in terms of a design? Well it's sort of a hub and spoke model that reaches out to multiple different groups depending on what the use case is: Pharmacy, the Institute of Precision Medicine as mentioned previously, CTSI, all sort of around the infrastructure for it and then UPMC with the clinical governance with testing involving medical genetics, HER teams, enterprise analytics and the health plan specifically. Our first deployment as I mentioned for this use case is within cardiology and it's focused on antiplatelet use after PCI.
These are the metrics that are collected. It is a clinical deployment, which means any patient that comes into Presby on their post procedure cath for post-PCI order set whether it's radial or femoral there is a pre-check genetic test on there. From that we have IRB approval to both capture those data on all patients and then to also approach those patients for a research study to see if they would be interested in sharing their long term data with us, bio-bank samples as well as persistent linkage to the medical record. So it's essentially an overlay that will come on top of the clinical recommendations. We certainly collect clinical outcome data but the overall hypothesis is that we can actually do it. When we started this project about 2 1/2 years ago now we had no HER processes, we had no testing onsite, we had no clinical governance so we were quite far away to make sure we could actually tests the process efficiency.
Here is our algorithm. So a patient comes in the cath lab, as I mentioned if testing is not done it is done. When the results come back to the medical record the patients are binned into these phenotypic categories, poor, intermediate, extensive and ultrarapid. These are sort of the clear examples but there is more alleles underneath them. And they are recommended to either stay on therapy with Clopidogrel or essentially move up to a higher potency agent, either Ticagrelor or Prasugrel. In the absence of other clinical factors of course that would drive the drug selection. It's important to know that of course genetics is only part of the story and the clinical factors often do override. So in drug interactions, high bleeding risks, other things may impact whether we change or not.
So Aim 1 was actually the governance, we created a clinical service and we created those protocols, hired the folks and started the teams to be able to do that. Aim 2 was to create testing, so this was a new onsite test developed down at Magee. It was validated under CLIA and was made available to our patients. This is the test report coming back, this is a paper based report. I'll show you the discreet results later, how we do decision support. The final step here is informatics and analytics.
So when we started this was what we had. We had a great cath lab registry, we had obviously the clinical care going on and we had several great HER systems. In order to get to where we needed to we needed to add this, we needed to add a lab, we needed to integrate our Enterprise Data Warehouse and we needed to create some sort of educational knowledge base to be able to help this. So everything in blue here is what we've had to create over the last 2 years. So the policies, the alerts, the decision support, established framework for releasing results into the medical record, multiple medical records unfortunately, get the genetic data in the right format to store it and then the overall goal was to really determine whether we have an increased value for our patients.
This is an example of how it looks like within one of the systems. You know this is - you can see the discreet results here in Epic and you can see if you click on that you can get the full result but all of our result reporting and alerting comes off the actual phenotype in the medical record, so we don't have to go through paper based reports for lifelong results.
In terms of the clinical service our team writes notes on every patient that's been tested, they are quick, they are simple, a couple of lines here. In this case the patient was already on therapy so Solomon Adams who is a graduate student and a pharmacist in our lab actually back over there is the one that wrote this note, and essentially the patient was already on Ticagrelor when intermediate metabolizer phenotype came back for the recommendation was to stay on current therapy that essentially we already had picked appropriately for other reasons beforehand. Again contact information, if anybody has any questions that's our pager number at UPMC and it's on the bottom of every report so we are able to query the consult service.
Alerts are in the medical record, so this is a best practice alert in Epic as an example and you can see it fires in a downstream order requisition for the same patient in the future whether it's 2, 3 you know 5 years in the future in any of the EHR systems we get the same alert that's fired in order to alert downstream providers that perhaps we should consider alternative therapy.
On the research side of things I'll go really, really quick through this because it's probably not the focus of this, and I think Dr. McNamara will get to it much more later on, but there is a simple consent process that allows those elements that I mentioned before so we can delve into the medical record, we can recontact patients at 7 days, 30 days, 6 months and 12 months to collect outcome data. We bio-bank and we have the capacity to share data across the network.
So we are a part of the IGNITE network in CPEC, the group that writes those guidelines and we have started prospectively sharing data with the institutions you can see here and probably guess what most of them are; but importantly allows us to have strength in numbers. We have about 5,000 patients so far in this network and we'll have our clinical outcomes being published very soon, about a month and a half from now to be able to share to the world what it looks like when you do these pragmatical clinical trials in multiple groups.
Data thus far, this is just sort of the lab test data because we don’t have our clinical outcome generated at 12 months yet for our own patients. But what's exciting here is you know essentially the frequencies are what we expect, so intermediate loss of function variants are 29% in our patients currently. We are about 550 patients or so as of last month, or the end of - or early last month. We've reduced our turnaround time in testing from 5 to 7 days to 27 hours so we can get it much lower than 24 with increased folks in the lab. And our actual variant frequency is about 21%. So 1 in 5 of our patients would be suggested to have some sort of change in therapy.
So I'll stop there so that Dr. McNamara has appropriate time as well. But I would argue form a pharmacogenomics standpoint this is just a simple use case, there are many others and many others will be deploying at UPMC. And we understand genetic variation is a source of variability in kinetics and dynamics, that this particular gene pair is a leading example that's being deployed at a handful of insouciants across the country. UPMC is a lead adopter and a leading driver for research in this and to be able to lead clinical care and certainly in the Pittsburgh region but also beyond. Several institutions have gone beyond this should we test thing and really thought about what do we do when we have the data, which I would argue is the appropriate approach. And there are certainly many barriers out there but I would say that it's not coming - well I can say it's not coming now, finally after a couple of years of doing these types of presentations, it's here and it's been here for about 9 months and the expectations is it will expand to the other cath labs in the system this fall.
So this is the main drivers obviously of creating the project and you want to follow the status of us, we are sort of out there on social media as well I guess if you want to do that. But it's certainly not me, it's a large team and this is just some of the groups that are involved and some of the funding that's involved. We do have significant support.
And while we are transitioning I'll just leave you with this picture, a little cartoon I got permission for of course with the copyright folks but I do think we'll get to this and hopefully this won't be a dazed pharmacist looking scared, but I do believe we'll be putting these stickers on the side of bottles very shortly that says you know look there is genetics guided dosing and perhaps we should think about it.