Ask AI
Back Back
Screening for DPYD Variants: Practical Guidance for Your Practice in my Candid Conversation With Dr Howard McLeod
Activity Information Ask a Question

Activity

Progress
1
Course Completed
Activity Information

Released: December 15, 2025

John Marshall
John Marshall, MD
Catch Up On All Episodes of Oncology Unscripted

Watch Now

Can Pharmacogenomic Testing Make Oncology Safer? A Candid Conversation with Dr Howard McLeod

 

John Marshall, MD:
Hey, everybody out there, John Marshall for Oncology Unscripted. You might notice that my studio is a little bit more bare today. One of the reasons is that I’m moving. They’re kicking me out of this office for a smaller office. That’s what you get when you get old. It’s discrimination—age discrimination. It’s actually job-appropriate job discrimination.

 

But we’ve got some blank walls in here, and we’ve got a lot of bright content, and I am so excited to introduce to you all—if you don’t already know him, pretty famous guy in his own right—Dr. Howard McLeod.

 

Howard and I have known each other for decades. We can say that, I think, with some certainty. And I want him to introduce himself, because what we’ve got to talk about is critical to almost everybody’s practice who’s listening in.

 

Howard, welcome. Give us a little background on who you are and what your area is.

 

Howard McLeod, PharmD:
Thank you, John. It’s great to be on your program. I’m an avid listener, first-time appearer, as they say.

 

I got dragged into this area of pharmacogenomics the hard way. A little girl nearly died of her therapy instead of her leukemia when I was a fellow at St Jude’s. We ended up finding the gene responsible for that.

 

Then a similar situation when I was over in Glasgow—a man got one dose of 5-fluorouracil and nearly died. So genetics as a way of predicting toxicity has been the theme of my life going forward—lots of implementation work, lots of large clinical trials and such.

 

Fast forward to our work at Utah Tech University: a lot of what we’re doing now is asking, “What do we do with this? How do we make it work in the real world?” Making it work at Georgetown is one thing. Making it work at a small practice “in the wild” is a totally different thing.

 

So, how do we implement this in all places, as opposed to just a few big centers? And the dihydropyrimidine dehydrogenase story that we’re going to talk about today is just one example. There’s a lot of activity around supportive care—Which pain med? Which antiemetic? Which antidepressant?—but this specific example is very much a toxicity example, and I’m looking forward to discussing it with you.

 

John Marshall, MD:
Let’s drill into this. One of our themes for this season right now is the increased utilization of molecular profiling—identifying positive signals for therapies.

 

I’ve always thought that pharmacogenomics—because I work with a guy named John Deeken here, who is a you, John—yeah, and he’s been preaching your same gospel for a long time: that we need to be doing more pharmacogenomic testing on our patients to optimize dosing.

 

There’s another discussion going on within the cancer community that we’re probably overdosing people, even up to the target, much less to toxicity.

 

Having lived this and preached this a long time, give me your take: why don’t you think pharmacogenomics has had stronger uptake in the community compared to, say, tumor molecular profiling?

 

Howard McLeod, PharmD:
Yeah, the oncology community is influenced by a couple of key things. One is strong, large, randomized clinical trials. It’s a little more difficult to do those kinds of trials for individualizing therapy—though they have now been done—for individualizing classes of therapy, not just individual drugs, particularly in the supportive care space.

 

So that’s part of it. Reimbursement needs to be there. There has to be a way of getting the test with a rapid turnaround. But part of it is also just: what are you used to?

 

I remember when PET scans first came out, no one would order one. Finally, radiology got more comfortable calling them—it wasn’t just nuclear medicine physicians, but radiologists would now call the PET component. Suddenly, you visit a cancer patient and you get a CT/PET. Maybe not quite that much, but conceptually it’s become so normal.

 

I think what we’re seeing now is a bit of a tipping point where some of the experience we’ve gained with tumor genomics and with inherited cancer genomics is causing us to say, “Wait a minute. I don’t need to waste my time—or the patient’s, more importantly—giving them the wrong drug when I can weed some things out and choose a better one.”

 

So, I think now it’s become more of a clinical efficiency tool, and that’s where we’re seeing the uptake start to happen.

 

John Marshall, MD:
There have been early adopters on DPD testing and then later adopters—I think our group here has been later adopters, to be quite blunt.

 

I’ll share with you that in the last, I would say, almost 10 years, when I write that first script for 5-FU or capecitabine for somebody, my sphincters are a little tight for the first couple of weeks. I personally have never written a script for somebody who turned out to be completely deficient at a life-threatening level, but I’ve taken care of several patients who’ve had that diagnosed and then come to us as a tertiary care referral.

 

Start us off, because a lot of our audience probably is not 100% sure where we are. I’m not asking for your hour-long, grand rounds, high-level talk, but: what are we talking about here? What are we testing for in these patients?

 

Howard McLeod, PharmD:
Yeah. So 5-fluorouracil has been around for longer than us—it’s been around a long time. Most of it is degraded to a less active or inactive metabolite that is then eliminated from the body.

 

Dihydropyrimidine dehydrogenase—or DPD, or DPYD as the gene name—is responsible for that first degradation step. If that enzyme is impaired—or completely deficient—then a lot more drug gets shunted down the active pathway: incorporating into DNA, into RNA, inhibiting thymidylate synthase, all the mechanisms we talk about in terms of how this drug really works.

 

So you end up with this preferential shunting into the active pathways. If you can’t get rid of it, then you’re forced to “use it,” and it all goes down that path.

 

Initially, the cases that were found by Bob Diasio and by other groups in the Netherlands and elsewhere were really those patients who had the extreme toxicity. Death, obviously, is as extreme as it gets. Then there’s prolonged, hospitalization.

 

The case that I first got involved with, where we ultimately discovered the DPYD*2A variant, involved a patient who had 21 days of ICU stay before he survived. His oncologist then asked for our help to figure out what in the world went wrong.

 

John Marshall, MD:
How many variants are there then? Are there… innumerable?

 

Howard McLeod, PharmD:
There probably are a lot more variants that haven’t yet been found, but currently the Association for Molecular Pathology—AMP, as it’s called—has two tiers of variants that they consider essential.

 

The first tier is especially essential, and there are 7 of those. In the second tier, there’s a lot of data but not as robust, and there are 7 of those as well. So, 14 total variants.

 

There have been a number of case reports that have come out even in the last couple of months, finding new variants. Those need to be vetted and proven to be truly pathogenic.

 

But the idea is: a handful or two—well, two handfuls or maybe three handfuls—of variants seem to be the ones that cause most of the difficulty in many populations.

 

So, for example, Jay Patel at Atrium Health in Charlotte has found some variants—and they were an early adopter, they’ve been routinely applying this—and he found variants that were more commonly seen in African American cancer patients. There have been others found more commonly in Southwest Asian patients—Indian, Pakistani, and so on.

 

But in general, those 14 seem to capture the bulk of it.

 

John Marshall, MD:
I was thinking about the situation in Europe, where there was a case—I think in France—where a patient died from toxicity. Correct me if I’m wrong, but it’s now a law in France, not all of the EU, that you have to measure this before you give 5-FU.

 

Early adopters have said, “Any information is better than none.” The later adopters, to your point about us continuing to find new variants, say, “I’m not really sure what to do with the result when I get it back. Does it mean I can’t give the drug, or I need to modify it? And if so, how?”

 

That lack of tightness, if you will, has been one of my personal struggles. But do you think now we’ve got enough data that we should be doing this in everybody?

 

Howard McLeod, PharmD:
I do. I’m obviously biased toward the early-adoption end of things, but I think we now have the ability to get fairly quick turnaround testing—not same day, by any means, but within 3 to 5 days. It’s being paid for by more and more insurance companies, so that issue is improving.

 

What we’re seeing is that if you have two bad copies—if you’re homozygous for variants—that’s really an extreme event.

 

Alan Venook at a presentation to the FDA back in January, during a public forum, data from a large Mayo study were shown in which the homozygous patients were the ones who did extremely poorly. That led to the first black-box warning in January from the FDA.

 

The heterozygotes had a lot of toxicity—hospitalizations, etc.—but they lived.

 

Those of us in oncology who are my age were literally taught, “You need to nearly kill the patient in order to kill the cancer.” So it’s difficult for folks like myself to look at this and say, “How can I under-dose this? Underdosing is worse than overdosing.”

 

But now I think we can refine it. We see that we can dose-reduce. If it’s an adjuvant patient, the recommended range is a 25% to 50% dose reduction. If it’s adjuvant therapy, I think a 50% reduction with subsequent dose escalation as tolerated is reasonable—and that’s a key point: don’t neglect the part about escalating as tolerated. If it’s a patient receiving active therapy for metastatic disease, then maybe you do a smaller reduction. Some patients will still get toxicity, but it’s unlikely to be fatal toxicity in that setting. So I think we’re at a point now where we’re as good as we are for renal dysfunction or liver dysfunction. It’s far from a perfect world, but we know this patient needs to be handled slightly differently.

 

John Marshall, MD:
That’s a really good point. I’m sitting here thinking about my electronic medical record, which I love and hate. The nurse brightly calls me to say, “Do you know the creatinine is 1.6?” And I say, “So what?”

 

What do I do? I probably do nothing most of the time, but every now and then I do something. There is a pop-up in my EMR that flags this, though.

 

But there is no pop-up in my EMR asking, “Did you do this [DPYD] testing?” Even though I’ve got all these criteria built in for other drugs.

 

I’m assuming that’s the next evolution—we’ll begin to increasingly see that kind of analysis.

 

Howard McLeod, PharmD:
Yeah. Dr Max Smith and Dr Jay Patel, whom I’ve already mentioned, and a few others have been working hard on that at their own institutions.

 

Right now, it’s mainly places with champions who are doing the hard work of integrating this into the EMR. But now that there have been two FDA label changes in the last seven months, and now that there’s an NCCN guideline change—at least for colorectal cancer, not yet for breast—places that make more wholesale updates are going to start doing that.

 

So Epic and Cerner and the others will start making big changes. The lookup tables that we all use electronically have now been updated.

 

It’s making it so that someone who doesn’t really know how to pronounce dihydropyrimidine dehydrogenase can still benefit from this. Frankly, you don’t want to know a lot about the enzyme, you just want to know, “I ordered the test.”

 

There was a day before your and my time when measuring bilirubin was a big deal. It certainly hasn’t been for the last 30–40 years. I think there will be a time when measuring this stuff is just what we do.

 

But right now we’re in this transition phase. We don’t want to underdose patients any more than we want to harm them with toxicity. Figuring out which side we err on as we move forward is the challenge each oncologist is facing.

 

John Marshall, MD:
Remind me—what percent of our patients would have some abnormality in one or more of these genes?

 

Howard McLeod, PharmD:
It’s right around 5% that would have some abnormality. The extreme toxicity is very rare—it’s less than 1 in 1,000. You may encounter it in your lifetime. In a busy practice like yours, it might have come through, but most practices will never see that extreme, fatal toxicity.

 

The fatal part will be caught by the testing we’re doing now, but really most of the testing is identifying those people who are going to have grade 4 toxicity with hospitalization, and asking whether we can at least manage them.

 

There is an approach whereby, if you know a patient is heterozygous—they have one good copy and one bad copy—you monitor them more closely and act quickly on any toxicity. The onus is on acting on it.

 

There is an antidote available. It has to be administered in a narrow window, and that’s the tricky part—can we orchestrate that? Often it’s just easier, in practical application, to start with a dose reduction and then move up.

 

John Marshall, MD:
I’m just setting out to lead a national platform clinical trial in colorectal cancer where we have not included testing as an inclusion criterion. Should we add it?

 

Howard McLeod, PharmD:
I would, only because it’s now in the label.

 

Many studies will have done DNA collection and can test retrospectively. That’s certainly another viable option—prospectively accrue, retrospectively test. But if the trial hasn’t launched already, it would be a straightforward thing to add. There is cost and logistics, but there’s a lot of pretesting going on in a clinical trial anyway. You can send all the samples to the same place—Labcorp, Quest, the big labs all do this testing now. It’s not as hard as it was even five years ago.

 

John Marshall, MD:
Even the molecular profiling companies for their liquid biopsies have started to bundle this in, along with germline testing and so forth. It’s going to be in our faces whether we want it or not, if we order these other tests.

 

Howard McLeod, PharmD:
This is just the kind of thing we can do to learn more about the patient. We’re now at a point where therapy is so successful that we can care about toxicity, pain control, and all these other issues in a more nuanced way. We’ve always cared, but there’s always been this existential piece of “saving the patient.” Now that therapies are active enough, we can start putting more attention on this next layer.

 

We saw what happened in pediatric oncology a couple decades ago, where survival rates got to the point where they could worry more about cardiac toxicity and such. Breast cancer has had that privilege; colorectal cancer is now starting to have that privilege, where the therapies are active enough that we can start worrying about this next layer of issues.

 

So it’s an exciting time. It’s inconvenient that we now need to add more testing, but I think it’s an exciting time—and a very straightforward way to apply it.

 

John Marshall, MD:
And hopefully it makes us more successful. “Who needs it, what are the right doses, and how do we move forward to optimize outcomes for our patients?”—that’s what we’re all about.

 

Dr. Howard McLeod, thank you for jumping on and letting me borrow some of your time. I know everyone who’s listening in has learned a lot and probably tomorrow will start changing their practice.

 

Dr. Howard McLeod, thanks a lot.

 

Howard McLeod, PharmD:
Wonderful to be here.

 

This transcript has been edited for clarity.