AI CKD Test Clinical Implementation and Provider Pearls
Robert Busch, MD, provides his approach on implementing artificial intelligence (AI) testing for chronic kidney disease and offers success stories of this testing and treatment approach.
Transcript
Dhiren Patel, PharmD: For our viewers, we have a lot of [people who] may not be early adopters, but then once they see it here, there’s obviously a very tangible benefit; you do this once and you have a head start for your patient. But what does that current process look like? Can you walk us through the ordering of it? Is this a regular test? Talk to us a little bit about that current process and where you store this patient information as it relates to the pertinent lab values.
Robert Busch, MD: When I go through my labs, the patients who I’m seeing the next day, I look to see who has a urine microalbumin over 30 or a GFR [glomerular filtration rate],below 60 and that’s who [I test], if they haven’t had the test already. It’s a one-time test and it gives you that prognosis. It’s like ordering a cardiac calcium score in a patient that the cardiologist does all the time and says, “You need aggressive lipid management or beyond that,” and “You don’t need it as aggressively, even though they’re on a statin.” [It’s] similar here. If that one-time test is elevated, I look to see and I put it with my regular orders. Now in our practice we could send out the test to the company. Hopefully, eventually, this will be ordered in a regular lab that you put down KidneyIntelX and put the reason you’re ordering it, and it will be a send-out test, and I’m sure this will change over time as this becomes ordered more, but there are certain factors in the electronic record that the company has to have in order to run the test. It’s not just TNF [tumor necrosis factor] 1, 2 and KIM [kidney injury marker]. It’s what the urine microalbumin is, blood pressure, other clinical factors. So someone would have to have access to the chart to send that along with the test information so when they run the test, they could come up; the computer does its calculation and gives you the score. So eventually I would think commercial labs will have it. I don’t know if that’s where that is in the United States now. I know we ordered it in our practice and send it out with that EHR [electronic health record] data.
Dhiren Patel, PharmD: So from what I’m understanding, you have to have some additional inputs.
Robert Busch, MD: Some additional input that my study person sends out…. The phlebotomist draws the blood, and then the test comes back within a week. Then when I get that back, I counsel the patient [on] the score: “Good, you’re on enough medicine now,” or “You know what, I’d like to be a little more aggressive and add such and such drug.”
Dhiren Patel, PharmD: So it seems like you’ve incorporated the process and the logistics of it pretty well into your clinical practice. What are some potential challenges that you foresee and the implementation with this [for] your peers, where, I would think the primary care [providers]—those who are on the front line—this is something that they should be utilizing in pretty much every patient. But what are some things that you can probably see that might come up with some recommendations for them?
Robert Busch, MD: Well, where’s the blood drawn? If it’s drawn in a commercial lab, will the commercial lab have access to their data from the electronic record so that they could send that off to the lab to run the test? So the commercial lab probably won’t run a [KidneyIntelX test]. Renalytix, the company, runs the KidneyIntelX test, but they’ll need that chart information. So if it’s an office that draws blood in their lab already, then that’s a no-brainer because they have the electronic record data. If they don’t draw blood in their office, and blood is drawn from a commercial lab, will the commercial lab have special access with HIPAA [Health Insurance Portability and Accountability Act] and everything else there, to get that data that they need, or does the physician have to fill that out on the sheet ordering the test? I don’t know if that will be an obstacle for the physician or someone, a medical assistant, filling out the other clinical data on the sheet before the test is run. They need that to come up with the calculation.
Dhiren Patel, PharmD: Understood. Thank you so much. So Dr Busch, you’ve used this test now quite often in a good portion of your patients. What are some early success cases or patient use cases that you can talk to us about?
Robert Busch, MD: Suppose I had a patient with a GFR is over 60. So either stage 2 or stage 1 chronic kidney disease. And the urine microalbumin is in the mid-30s. So not that high, but it’s high. How do I know that patient is going to go on to worsening kidney disease vs…they have data that are there, but if I’m aggressive with their diabetes anyway and give them a RAS blocker, that they’re OK? And suppose their A1C is 75. So I want to do something more than the metformin. So what are my options? Well, I want to use one of the newer drugs that you’ll lose weight with and lowers A1C without hypoglycemia. So I might use a GLP1 [[glucagon-like peptide 1] or SGLT2 [sodium-glucose cotransporter 2]. Currently SGLT2 [has a] known renal benefit. Maybe GLP1s will have the renal benefit, but currently that’s not known. So I have the option of either one. And if it’s an overweight patient, I might be tempted for the GLP1, but their patient has a high urine microalbumin. Of course, you could say, “Well, give both of them,” and often I do that. But suppose their drug plan—they don’t want to get 2 branded drugs or something. If the kidney and TELEX test was 90-something and they were high risk, I definitely would give the SGLT2 that has known benefit, make sure they’re on maximal RAS blocker, or maybe give finerenone on top of that, which are all proven renal benefit drugs, vs if the KidneyIntelX test were in the teens, that’s not something I’m overly worried about. It would be if a cardiac calcium score was 0, I’d be less aggressive with the statin. If the cardiac calcium score were 1000, I would give evolocumab and ezetimibe to get the LDL [low-density lipoprotein] way down. So it’s the same kind of push therapy-wise. So that would be more reassuring if it was a low number. And in the studies, it showed a low number didn’t go on. They were 15 times higher to advance to end-stage renal disease or drop your GFR by 40% if your KidneyIntelX were 90 vs if it was low, vs someone else [with] the same GFR, same urine microalbumin, which is 1 point in time, but their KidneyIntelX test showed 15 vs 95—then I might go with drugs that, if [the person were thin], ideal weight, I might give a DPP4—where I don’t have to, it doesn’t have renal protection or heart protection, but they didn’t have heart disease, vs if they had the score of 90, I’d give the SGLT2.
Dhiren Patel, PharmD: So it’s a very useful test when you think about it from a patient and a clinical therapy segmentation.
Robert Busch, MD: Exactly.
Transcript was AI-generated and edited for clarity.
