Marla Dubinsky, MD, on Biomarkers in IBD: Finding Precision

Dr Dubinsky reviews her presentation from the Advances in Inflammatory Bowel Diseases 2021 virtual regional meeting on March 6 on how biomarkers can provide greater precision in choosing therapies and dosing regimens for patients with IBD.


Marla Dubinsky, MD, is a professor and codirector of the Susan and Leonard Feinstein IBD Center at the Icahn School of Medicine at Mount Sinai in New York City.


Hello, I'm Dr. Marla Dubinsky, and I had the pleasure of speaking today at the Advances in Inflammatory Bowel Disease meeting on biomarkers for the optimization of IBD finding precision.

The role of biomarkers and how well it's going to impact our practice and help us through a practice can be thought about in a spectrum of 3 distinct categories. One is prognostic biomarkers, which speaks to the ability to predict disease course.

Then the second grouping is predictive biomarkers, which I think about is predicting treatment response, meaning are there certain biomarkers in a particular patient that may help us understand how a patient is going to respond to a certain therapy.

I'm hoping at the end of the day we'll find some biomarkers that say, "If you have this biomarker, you shouldn't even be using this therapy, you should be using that therapy,” which would be exciting and heading us into a space of precision health and precision medicine.

The third category is monitoring biomarkers, meaning those that can monitor the disease activity and that are sensitive to change with treatment. When we think about predictive biomarkers, it's about deciding where are we in our proactive approach to IBD.

I think about it— and I've been studying the role of prognostic markers for a long time now— I think about it is that there are clinical markers that we know that are associated with worse clinical outcomes, such as, if you have a complicated phenotype in Crohn's disease, pancolitis, or deep ulcers with ulcerative colitis, patients who been hospitalized through independent biologic failures, these are patients that represent severity.

We've moved beyond clinical in saying, "Why do we wait for a patient to have bad prognosis for us to say, this is a more severe phenotype? Are there blood markers, stool markers, or genetic markers that may help us determine prognosis?"

If you think about where we've been, genetics has been shown, maybe a predictor of more stricture in small bowel disease serologies, which is what I've been studying for most of my career, these antimicrobial antibodies.

More recently, with the work that we've done, myself, Corey Siegel, and Laurie Siegel. We've put together somewhat of a Decision Support Tool using an individual's personal biomarkers of both serologies genetics as well as their clinical variables.

Instead of thinking them in silos, putting them in one model, which is now called CD Path. It's in a collaboration between Prometheus Labs and Takeda, where they're going to be offering this Decision Support Tool, which will help you predict the prognosis of your Crohn's patient, which is really exciting.

We're looking forward to helping that. That improves the way we're able to even share our prognostic predictions in a way with our patients to say, "You know what, even though you may feel good today, which is disease activity, there are markers that suggest that you have a more severe prognosis.

I'd like to intervene effectively and early because that's when I have the biggest chance of success. Now the question is which treatment?

Once you've decided that you have a patient is at high risk for complicating IBD and bowel wall progression—both UC and Crohn's disease by the way—how do I decide which treatment I'm going to take? There is a couple of ways we're thinking about that.

We may think about it as maybe disease location, which is not quite a biomarker, but I'm going to throw that in there to say, disease location. There are certain therapies that work may be better for patients with small bowel disease versus patients who have large bowel disease.

There are patients who have, for example, some genetic risk factors, for example, TPMT is something we've been using from a safety perspective. It doesn't have to deal with, you should respond to this class or that class, but it should tell us that if you have TPMT levels that are homozygous deficient, for example, higher rates of leukopenia, there is a new gene called NUDT15 variant, which has been associated also with leukopenia. When you combine TPMT with NUDT15, you're looking at explaining about 50% of the leukopenia. There are some labs that offer that.

Probably one of the more exciting things that are going to be coming out is the risk immune profile from Prometheus, which is the HLA haplotype or genotype that is going to tell us whether our patients at risk of developing antidrug antibodies against the TNF, specifically against infliximab.

Maybe that would help us decide if this a patient I have used optimized monotherapy, for example, or a patient that I would use combination therapy, but if so, do I need to look at the NUDT15 on the TPMT to make sure that they're not someone who is at high risk of leukopenia.

We're going to get better when it comes to also predicting response, but also looking at safety of these therapies.

In terms of predicting a response, the future is going to be either proteomics or gene expression markers that you may be able to measure in the blood, but right now, there is some studies looking at the mucosa of patients, for example, in some of the anti‑TNF data from the Fiona Powrie group in Oxford, showing that the oncostatin M expression may be elevated in patients who are altered and elevated in patients who go on to have non‑response to the anti‑TNF.

The question always is, are these severe disease phenotypes or is it actual marker of nonresponse anti‑TNF that needs further exploration?

There is some work that's been done looking at the alpha(4)beta(7) in the mucosa expression as a way of saying that infliximab may work better in patients who have a certain level of alpha(4)beta(7) expression.

The future is exciting to think about, "Can we measure it in the lining, but then can we actually measure it in the blood," because that's an easier way to determine treatment response when it comes to biomarkers. That's something we're following closely.

The third bucket, which is monitoring biomarkers, which is again something that I'm obsessing around these days, around the idea of what I'm calling and what we all refer to as very tight control.

The ability to proactively predict that your patient is not responding, rather than waiting for symptoms to tell you that or even needing to wait for mucosal evaluation. There are biomarkers such as CRP and fecal calprotectin that we know we've been monitoring treatment response with.

There are studies that have shown that even early on at the 8‑week mark in some of the clinical trials, if you see already a calprotectin less than 250, for example, that leads to better durability out to the year mark, as well as thinking about the way that we can integrate maybe our therapeutic drug monitoring as what we call a monitoring biomarker.

Meaning can we forget about being reactive, can we stop doing empiric dose escalation, particularly when it comes to anti‑TNF, and where we have the most data which is proactively checking a drug concentration early. When I say early, and it's beyond reactive, proactive means a patient is in theory doing well. You're proactively checking a level and you're going to do something about it, not based on symptoms. Historically, we've been communicating that at week 14, meaning the first maintenance infusion of infliximab, may be the best time to get a level.

Then we realize that that may be too late, because probably the magic is all about understanding which patients cannot make it 8 weeks from their third induction until their first maintenance.

There is also some exciting pharmacokinetic models that will hopefully be available soon as well, where you're going to be able to, just like the CD Path model I mentioned, where we can throw in a whole bunch of personalized variables for that patient.

Imagine the ability to throw in all of the pharmacokinetic parameters that influence drug concentrations in patients. You'll be able to show you when a patient needs to come into the day and to the milligram dose total, in order to get that ideal trough concentration that we know is associated with good outcomes.

This is an extremely exciting time in understanding the role of biomarkers. I think that as we move into this very proactive IBD management strategy, we're not going to be able to do it, unless we start to deploy objective evidence of ongoing inflammation.

Being able to predict which patients need to step up to a more effective therapeutic intervention. Keep them in tight control and get to your target of mucosal healing. Thank you.