Day 2 Recap: ACR Convergence 2026

Hi everyone, it's David Lou here again, hosting the RheumNow ACR Convergence 2026 daily recap for day two. We're really in the midst of a very impactful and busy meeting. It's been a lot of fun and I'm joined by another three friends here today to talk through some very interesting happenings for the day. So I'm going to go round the screen and introduce yourselves and where you're from. Anthony. Hello. I'm Anthony Chan, a consultant rheumatologist from Reading, United Kingdom. Eric. Eric Dian, I'm a rheumatologist practicing in Summit, New Jersey, here in Orlando. And Adella. Hi, I'm Adela Castro from University of Tennessee Memphis.

Fantastic. So some really interesting stuff out there today. Eric, I think we'll start with you. What caught your attention, caught your eye, big changing kind of abstracts? Yes, definitely. Some interesting sessions at the plenary today. I'm going to chat about abstract 1820 with what I found somewhat surprising results. So I'm interested to hear what people think. I thought very thought-provoking. This was looking at cardiovascular risk with gout. Something we've heard about in prior meetings. This was the post-hoc analysis of the CARES trial which looked at the association of gout flares with subsequent cardiovascular disease. A big trial of over 6,000 patients starting on urate lowering therapy.

So taking a step back, we know gout is associated with metabolic syndrome. We know that there is mortality cardiovascular risk associated with a gout diagnosis. There was a prior JAMA study in 2022 which showed an increased mortality rate of MACE after gout flare. I think it's something that gets quoted a lot — that there's an odds ratio of almost two in the 60 days after a gout flare, and that extends out to 120 days after a gout flare, that your risk of cardiovascular disease is elevated.

This was looking at similar outcomes but from a little bit different trial design. So this was looking at this large population of patients started on either febuxostat or allopurinol — all patients with established cardiovascular disease, or I'm sorry, established risk factors for cardiovascular disease — and looking to see do they have gout flares and most importantly were there MACE events. So this study differed from the prior study. The prior study only defined a gout flare retrospectively, looking at whether there was an ICD diagnosis code and were they prescribed therapy including colchicine, NSAIDs, and/or corticosteroids. This one had a little bit more detailed analysis and prospective following of the patients. They looked at both participant-reported flare or something called the GAFFO criteria. The GAFFO criteria includes three out of four criteria that are needed, which includes a patient-defined flare, pain of at least three out of ten at rest, a swollen joint, or a warm joint. So three of those need to be there. So it's a little bit more definitive of a diagnosis and it's not based upon the medications.

This was a nested case-control study — this is the post-hoc analysis of that prior trial that was looking at urate lowering therapy initiation. And so this study did not show that there was any increased risk of cardiovascular disease MACE events after the patients had gout attacks. And they looked at it in all different ways. No difference after self-reported flare, no difference after the GAFFO criteria. All the sensitivity analyses looking at physician-adjudicated flares as definite or probable, or just definite — swelling, warmth — any way they sliced it, they saw no difference in MACE events. So this large study on initiation of urate lowering therapy showed that there was not an increased difference.

So what do we make of it? I'm interested in what the group has to say. It is more detailed information and it does make me think perhaps it wasn't really the flare itself. The last study was really based upon the medications — wasn't it more the NSAIDs and more the steroids that increase it? So my takeaway is number one, it is safe to start patients on urate lowering therapy with cardiovascular risk even if we know that has an increased risk of flare, and number two, again being cautious with the medications. But I think ultimately we need a little bit more data for this.

Yeah, and I mean this is really quite high-fidelity data. We're talking about data from CARES, a really well-run study. The people running this — I mean this is Hyon Choi today and colleagues from Massachusetts General Brigham — you know, Hyon Choi, Dan Solomon, Sara Tedeschi — real kind of people who you trust. If you're going to trust anyone in this world to analyze these kind of data, these are the people you trust. And to be able to kind of turn around that JAMA finding is pretty important. So maybe Eric, can you just go into a bit more detail as to — you were saying about the therapies from the previous analysis. NSAIDs, steroids, two things we know have a lot of cardiovascular potential there. How do you tease that out? And aren't they just going to be inherent to the way in which we deal with gout flares? I guess thinking about colchicine as well and where that sits.

Yeah, I'd be really interested to go back into the other study and look at whether the colchicine group was different from patients that were on NSAIDs or steroids. And I don't know — I'd have to look back to see what kind of analyses were done in that prior study. But definitely we know that there are risks associated with it. I've always been counseling patients that the inflammation of gout itself is a risk factor. And I think certainly chronic inflammation is a risk. But I think the question is, is a single gout attack that inflammation risk factor? And I do think it is important to go back and look at what's the risk — we know risks associated with one course of corticosteroids increase that risk, and how much of that was driving that other study, or the course of NSAIDs. So I think it would be really good to understand that data from the other one more and want to go back and look at that.

Yeah, absolutely. And it gets to the idea of trying to think about how we manage cardiovascular risk in gout in general, in gout patients. Adella, how do you see these data in terms of your practice, and does this change the way in which you approach that cardiovascular management piece for gout patients?

So I agree with Eric here. Definitely I want to know a little bit more about the details in terms of like the effects of colchicine, because we do know there's some evidence that can hint that colchicine may have some cardiovascular protective effect. So that would be interesting to see. I definitely am very aggressive with my gout patients because not only do they have gout, but gout is also a marker of metabolic disease. So definitely these patients have more than just one risk factor — diabetes, obesity, hypertension — so definitely those are factors that need to be managed appropriately, and of course we need to manage gout appropriately. So I tend to manage my patients very aggressively and try to minimize that flare. So definitely that's good to hear on this side, in terms of the patients that were on ULT probably had less, you know, major cardiovascular events.

Well, let's move on. Interesting stuff to take away to the clinic. I'm sure there's a lot more that we can do better with gout and we will see a little bit more later in this meeting which is going to be very interesting. But maybe I can throw to you now, Adella, about what you found interesting today. I think we're talking a bit about some guidelines that you saw.

So a lot of interesting stuff today. My favorite, which was very needed, was the 2026 ACR/ASAS SPARTAN guidelines on management for axial spondyloarthritis. I'm going to try to summarize the major hallmarks — kind of what are the things that they added this time. So in terms of assessment of suspected axSpA, of course they still recommend getting the history, the labs, and of course having those extra-axial manifestations in mind. Then for the first-line imaging they recommended specifically getting the anterior pelvis x-ray with an AP view — just a single view — as opposed to the previous recommendation of the Ferguson views, and then if that was negative the recommendation was to obtain the MRI. And this is something that is very high yield right now because they were very specific about the MRI optimal acquisition protocol, which guides a little bit more how to make the right diagnosis for axSpA on MRI. Something that was also interesting that they included in this guidance: if for some reason you do not have access to MRI, then you can actually use low-dose CT to diagnose patients with axSpA.

So once you have a diagnosis, then how do you manage? And I think this is still very controversial a little bit, but the recommendations this time included: if the patient had no risk factors for progression — meaning normal CRP, no syndesmophytes — then you can still consider NSAIDs as first line. And they were very specific — this time they included a recommendation on specifically COX-2 inhibitor NSAIDs as opposed to the other ones, and they of course recommend continuous NSAID use as opposed to intermittent or PRN. But if the patient had high risk for progression, then in this case you could start the patient on a TNF inhibitor in conjunction of course with other therapies that are very important, like physical therapy — which they also made a recommendation about specifically: they recommend aquatic therapy over land therapy.

And then of course, once you assess and decide to start the patient on a biologic, before they were recommending TNF first, but now there's no difference — you pick pretty much between TNF inhibitor or IL-17 inhibitor, whichever you have access to, and take into consideration what other comorbidities the patients have. And then something that also caught my attention for the first time — in terms of patients that are difficult to treat, because sometimes those were very vague in the previous guidelines — this time we have a bit more recommendation on difficult-to-treat patients, and this time they also recommended the evaluation of nociplastic pain and even some guidance on how to manage nociplastic pain. And then they also recommended — specifically this time — if the patient is really difficult to treat because of comorbidities or anything else, they do recommend combination biologics. So this was something that was also kind of a highlight, especially because there's limited evidence on combination biologics or combination targeted therapies, but the fact that they mentioned this in the guideline is also very helpful to kind of explore more on these. So I kind of tried to cut it down in a nutshell, but I want to hear your thoughts about it.

Oh, I mean you hit some very interesting points there. Thank you for cramming that all in. And fascinating that in these anticipated guidelines — and guidelines are usually fairly conservative — that all of a sudden we've had a swing toward combination biologics, which I guess speaks to how mainstream the idea is becoming. Anthony, you're my font of knowledge as far as the therapeutic approach is concerned. I turn to you often. So tell me a little bit — is this expected? Having TNF and IL-17 on equal footing is quite interesting. Are these guidelines more or less what you thought they were going to be?

I think we have now got the ASAS guidelines — I was involved in the UK BSR guidelines — and now we have the ACR one, and I think it's emerging evidence when these guidelines are written, what's available at a time. But some interesting things: they're going for more of a risk stratification approach — are you a fast progressor or low progressor — and then in the second phase, do you meet criteria for difficult to treat or difficult to manage? So it's a more individualized type approach, which I think is the right thing to do. They also allow choice between TNF, JAK, or IL-17 and even combination. So I think it's an evolving thing. The guidelines will evolve as the evidence evolves, and I think it's a kind of right approach.

The NSAID one is interesting because that's kind of come back again and we have this constant discussion whether on-demand versus continuous is better. We've been through that many times. But I think it's good that it's put out there as an option again. You know, for me it was — particularly the NSAID — these are usually young patients and not that compliant with pills, and you really want them to get into remission, so of course NSAIDs help quite a bit with the pain pretty quickly. So probably yes, but like I am a little bit hesitant on just keeping my patients on just NSAIDs even if they're low risk. I think I'm a little bit more nervous — if they don't take the medication — about how to monitor them closely.

Yeah, I think with these patients you have to monitor them, and the thing is, access to MRIs — how often can you do these things if you are putting somebody on a drug where the disease could potentially progress? You also mentioned low-dose CT — there was also a nice abstract today that showed it probably was a bit more sensitive than MRI in terms of picking up structural progression. Of course that's difficult because of the radiation involved if you use that as your measure, but there's no doubt it was better than the plain radiographs and perhaps even better than the MRI.

Eric, you've got a bit of interest in how this imaging plays out. Did this kind of surprise you, some of the recommendations, or do you think this is really the start of better phenotyping from imaging — is that a part of where we're going?

Yeah, I think phenotyping, understanding, really classifying again who's at risk, what their progression is, understanding these radiographic, non-radiographic — I was going to say just more generally, the guidelines help us decide how to practice clinically. It's also just useful for access. In the American system, often needing to justify — the idea that they put out guidelines saying TNF inhibitors and IL-17 can be considered equally, and particularly that combination biologic is a guideline-directed recommendation, I think can be really helpful just in terms of getting access to these medications. So that was actually the real-world practical part that I thought can really change our practice.

Something about it in particular that caught my attention in the guidance — which is kind of the hot topic of today — is how do you identify what is the MRI definition of axSpA? And then in the guidelines they were like, make sure that you use this protocol, but they don't mention the specific protocol, which I think is important — it needs to be kind of widespread. But again this protocol has its limitations because it was kind of made using central readers, but we already know we don't always have expert radiologists reading our MRIs. So we definitely have to kind of stay in contact with our radiologists, make sure we are asking specific questions.

There is a huge weighting now on imaging in terms of classification of axSpA because of the new ASAS SPARTAN guidelines, and I think this is why today at the plenary they also presented the SPARTAN/ASAS MRI indicative of axSpA — that's the new definition — where there's a bit more weighting on erosions compared to the old ASAS guidelines where it was more inflammation. So I think that's going to influence how we classify patients for clinical trials. But lots of posters around where they have found that it actually does exclude a lot of people in some countries where they don't have similar access to MRI.

Let me quickly throw in the one question — I feel like this has to be our first mention of AI for today, won't be our last. Is AI in imaging going to change a little bit of this? Is that going to help us to phenotype? All of a sudden we've got all these choices, we're changing all the guidelines — they were very boring for a long time and all of a sudden things are really front foot. Does that help to unlock a lot of our personalization?

I think there is this issue with standardization and central reading — that could be a role for AI, to adjudicate between different readers, where you have a standardized approach. And in places where you don't have specialized radiologists, that could potentially be a place where this can be done, but it still needs a human validation at the end of that.

Yeah. Oh, sorry, go ahead. I think again, I agree with what you said. Some places just don't have experienced readers and AI will definitely bring the experience of having looked at a lot of them. It will be interesting to see though, as AI gets better and better, will it be picking up these micro-erosions that are extremely sensitive — and what does that mean for clinical relevance? So it'll be interesting because as it gets better and better I wonder if this sensitivity might just produce more results that we then have to decide what to do with.

Yeah, I agree that imaging is important for diagnosis, specifically for our early patients, which is what we want to get them. And having a good MRI result that we can trust is going to be very important. So not only do the radiologists need to have the education, but also ourselves as rheumatologists who are in charge of actually making the diagnosis — it's going to be important to understand what exactly it is we're looking for, and if we don't have those results or we're not sure, we definitely want to communicate with our radiologists.

So before we move on, I just want to quickly ask — of the three of you, who's using combination biologics? Yeah. Yeah. Got one take. Yeah. Yeah. Yeah, I mean I think that's all — everyone here. So maybe it'd be interesting to hear thoughts on where people think this is going. Clearly it's not going to be for everyone. Clearly there are some people who will benefit. How do we pick the winners? How's it playing out in your practice?

So for me in particular, the most common situation where I find myself needing to use combination biologics is when I have patients with concomitant poorly controlled inflammatory bowel disease. And so far I've been doing very well with — as the evidence has been showing — the combination with the IL-23 inhibitors with the TNF. So far that has been quite a success. But sometimes I've also been needing to use JAK in combination with IL-23 or TNF as well, and of course we definitely need more data on safety, but so far the data that we have on safety seems pretty safe, so we'll have to see more.

Yeah, there was the study from the Spanish and the French groups here at ACR yesterday and today, which I think replicates what we've seen at previous ACR and EULAR, that generally it's quite safe. I think that's because of careful selection of these patients. And a lot of what we do is in combination with dermatologists or gastroenterologists — we're doing it within a combined clinic setting where we have decided that we would go for combination therapy. But I think it will become more common.

Fascinating how it's flipped from when we were in RA and TNF inhibitors and anakinra — that didn't work out — and then for so long we flipped away from this, it was completely prohibited on the label. And now I feel like the move in SpA has emboldened the rest of rheumatology. We haven't quite hit our safety barrier yet. It will be very interesting to see how it plays out.

Anthony, tell me a little bit about what you saw that was good on the floor today.

So I think this ACR is focused on obesity and BMI — it's heavily, if you can use the word, weighted on that. I was interested in abstract 135 and this is the bimekizumab data because it's a large dataset. What they did is they pooled the BE Optimal, BE Complete, and BE Vital. Now we know that in PsA, 40% of patients would be classified as obese or overweight. So in this study they had about a thousand patients who received 160 milligrams of bimekizumab every four weeks and they followed them up for three years. The ACR50 was then divided up by whether they had a BMI of less than 30 or more than 35. Those who had a BMI of greater than 35 — at one year the ACR50 was 41%, compared to the lighter group which achieved 57%. And at three years that number remained the same — it didn't change after three years.

Now we would expect that if you had achieved an ACR50 maybe you became more active, you could do more things, but in fact it didn't change in that time. The percentages remained the same — it was static over three years based on your initial BMI. And what's striking is the gap between those who had a BMI of 35 and those who had a BMI of 30 — there was somewhere between a 17 percentage point difference between the two. Now of course we will see the 52-week data from TIETAR PsA on the last day. But there is this whole — we knew this for a long time — that the higher your BMI, the poorer the response from biologics. But this one shows you that not only that, you stay within that same class even three years down the line. So this calls for another approach — maybe looking at GLP-1s in these people who have then achieved ACR50 — that the combination would be a metabolic approach to treating them in terms of getting them further down on their tender and swollen joint counts.

And I mean this is something we've always seen in our practice. I think now that we've got something that we can actively do about it — with more options to think about — all of a sudden I think we're starting to realize what we might be able to do. And I guess we're starting to be more concerned about where the causal relationship sits. So Anthony, just speculate a little bit here for me — it's a safe space. It's only the entire internet listening to you. Tell me how you think the causal relationships might sit here. We know we're trying to interpret things through post-hoc analyses or registration data not necessarily set up to answer these questions directly, but what's this relationship between weight and disease activity, pharmacotherapy for both things, and activity? How do you see it?

So what is quite clear from this data here today, and also from the TIETAR PsA data which will be presented later in the conference, is that treating inflammation alone is not enough for us to achieve better joint outcomes but also weight loss. So when these people achieve ACR50, their weight still remains — sometimes the reduction of the weight is not as significant as we think — that by getting them into good disease states they may exercise more. So there is clearly a metabolic pathway here. The incretins are doing something on adipocytes which we know are pro-inflammatory. So it's not just the weight that is causing joint loading here, but it's actually an inflammatory cascade that I think is at work. Because if you look at the TIETAR PsA data now — they didn't have a tirzepatide-alone arm in that study — but you could read from the combination versus the ixekizumab alone that there's a 12 percentage point improvement. So every 100 patients, there are 12 other people who improve in terms of their joints just being on the extra GLP-1. So that's where we are going. It's not just all inflammatory. And this brings up the whole thing about non-inflammatory issues that we are discussing about difficult to treat, difficult to manage — while we are trying to understand all of that, I think this is potentially an avenue for us to go and explore in terms of the non-inflammatory aspects of PsA.

Yeah, fascinating stuff. Let's move on — well actually sorry, does anyone have thoughts? I mean this is very live in our practices. Eric, thinking about how we're treating our PsA patients, approaching our PsA patients — has this changed the way that you counsel your patients and the way you approach this management? And I'm interested from Adela as well — very different parts of the country, different perspectives are really important.

Absolutely. I think, just like gout, PsA is also a metabolic disorder. And so it is — with hypertension, hyperlipidemia, diabetes, gout, PsA — I think they're all in that same umbrella. We should think about cardiovascular risk just as we talked about with the first topic. It all is under one umbrella. And I think we need to not just focus on the joints and the skin, we need to think about it globally. And number two, all this data — the TIETAR PsA and other studies — we now need to think about GLP-1 or just weight reduction as disease-modifying therapy. And I think that's true for osteoarthritis — we've seen that at this meeting — and for inflammatory arthritis, and for nearly all of our conditions. Addressing the weight is potentially disease-modifying for a lot of these conditions.

And Adela — that's probably pretty similar to your experience?

I agree, and the counseling is very important. I was actually talking to one of my colleagues when they came out of the session on GLP-1 and psoriatic arthritis, and they were saying — you know, counseling the patients — you may want to do it with kind of a grain of salt. You may want to ask permission from the patient if you want to counsel them on obesity. Which — how often do you ask permission to counsel your patients? So that was interesting to hear from that session. And I think it's important — and this is definitely something helpful to have — because we do have that modifiable risk factor that not only causes inflammation but for so long was related to poor outcomes in these patients. Now we do have a tool to modify this risk factor. And of course when we do see the weight loss, patients are a little bit more energetic, they have more ways to exercise. So that of course changes a little bit of everything — it not only helps with their actual inflammation, but also helps with, as Anthony was saying, the non-inflammatory part: the depression, the nociplastic pain, all of them.

Yeah, it's tricky, isn't it? I think it's easy to offend in these situations, and different people have very different perspectives. And if you get people offside at the first question, then you're not going to have much luck with the rest of the conversation. That opening is very important.

So I just want to talk a little bit about — in the same theme but then a very much a right turn — talking a little bit about what traditionally have been SpA-related therapies and how they're translating to PMR and GCA. And there are some really good reasons for that. We've been really focused on the Th1 axis previously, and hence tocilizumab, IL-6, and I guess when we've looked into IL-1-related therapies. But also thinking about things like sarilumab — where we know it's going to cover that — and then in recent times there's been a bit of a shift. Well, I mean, it's always been there, but I think there's been greater attention. Success breeds attention. There's been growing attention to IL-17/IL-23-focused therapies in the PMR and GCA space.

So in the PMR/GCA abstract session there were two really interesting abstracts on that. Firstly, I think it was follow-on data from the REPLENISH study. A lot of us will be aware that secukinumab in relapsing PMR came out earlier in the year. It's looking quite good. I think it's going to lead to real-world use in the very near future in multiple jurisdictions. And what this looked at — and I think this is always a question with possibly phasic disease, sometimes phasic disease like PMR, relapsing PMR, maybe less phasic — what happens when you stop the therapy?

In REPLENISH, after the first 52 weeks — last active therapy was at 48 weeks, but patients should still be on steady state at that point at 52 weeks — they stopped both arms, the placebo and the active therapy. What happened? Well, about 50% of people flared in that situation. So, shock horror, relapsing PMR is a chronic disease a lot of the time. But then does secukinumab give you a legacy effect? There's a delta you can see between the two curves — there's a difference — but really it's possibly just drug wearing off, maybe there's some limited legacy effect. I think we have to dig into that a little bit more. Of course ideally what we'd like to do is hit people with a therapy and have an effect that goes beyond, but we may be stuck in a situation where we're giving people secukinumab in the long term, much as we would for RA and axSpA and PsA. And once upon a time people thought the idea of giving ongoing therapy for PsA was ludicrous, and we've come a long way since then.

The second part of that is actually in the same vein — in GCA. And I remember some years ago watching our friend and colleague Richard Conway on the late-breaking plenary stage talking about his study looking at patients who received ixekizumab in GCA — looked very promising there. There was a small series from Mass General which looked a lot less promising. So in that context the French have done a phase 2 investigator-initiated study — because it's off patent — a randomized controlled trial, first-time proper RCT data. And what you can see is that it's slow to kick in. We know ixekizumab can be a bit slow to kick in, but there's definitely a wide gap between the placebo and the ixekizumab arms. So it's really interesting, especially in the context that the GUIDE study — gelamumab, the IL-23p19 inhibitor — in GCA was a bit of a bust earlier this year. So I guess there are all sorts of questions. Is IL-12 doing the heavy lifting? Is there something different about this? Is it partially about the study design? And will a phase three investigator-initiated study actually happen?

So let me ask all three of you — first of all, are any of you using biologics at all for PMR? And I guess this is a thing now — we've obviously got some options, we've got sarilumab in the US. And I love this because we also had a great lecture from Mike Putman of course about this, and there's still the particularly the REPLENISH and the SAPHYR trial. We're all in relapsing disease, you know, so technically you kind of have difficulty tapering the steroids. But in my clinical practice, these are patients that are already having so many comorbidities — diabetes, very high risk for osteoporosis — and do I want to wait to taper down the steroids that long? So having access to a biologic and having the option of starting a biologic early and decreasing the steroid burden for me is life-changing, and also for my patients. At the same time, having more than one option — because I've had that patient where the patient did have a side effect to sarilumab and then I was like, what can I do now? But now we do have other options that I can start my patients on. So definitely having that option of biologics for these patients is great.

Yeah, and I think it is entirely possible we might have more options in the future in relapsing PMR. There's been a couple of things thrown around at this meeting. And I guess getting to your point as well — newly diagnosed, if we can get in there early. Who knows — maybe it's all to do with — yesterday we talked about generic tocilizumab, sarilumab, and getting it from Cost Plus Drugs and all the rest. Who knows what that might do. There's obviously some data there.

Great. So maybe we'll just quickly — time is getting upon us — let's do a quick-fire round around the group about something interesting that you saw out there. I'll start with you, Eric.

Yeah. So yesterday you talked a lot about some of the new mechanisms for myositis and particularly dermatomyositis. But I think to zoom in on upadacitinib — which is now FDA approved in the US, that is something that is available for us now, the TYK2 JAK inhibitor — there was a range of different posters on it, 1584 to 1587, and just a few quick hits from those. One of them looked at skin improvements beginning at week four, improvements of CDASI itch scores, skin-related quality of life. Another one on steroids — looking at the subgroup, three-quarters of the patients entered the study on steroids — those patients did well, they were able to taper the steroids. They looked at the subgroup that had interstitial lung disease — one-fifth of them had baseline ILD — they didn't specifically look at any endpoints, so you can't comment on that, but it seemed to be safe in ILD patients. There wasn't any kind of signal for ILD progression. So I think definitely more information is needed there. And then the fourth poster was looking at patient-related outcomes, and the PROs looked favorable for functional disability, pain, and fatigue. So I thought that was useful data for something that is now on the market for us.

Yeah, very exciting. Adela?

So I am going to go on that same note of the hot topic of GLP-1. [laughter] So this is abstract 1898. In this study, the objective was to evaluate the association between GLP-1 receptor agonists in patients with rheumatoid arthritis that were on DMARDs, particularly on methotrexate and TNF inhibitors, and they wanted to evaluate the risk of malignancy at five years. The specific malignancies that were high yield here were lung cancer and non-Hodgkin lymphoma. So this was a retrospective study using the Trinetx database including over 24,000 patients with rheumatoid arthritis, and they did propensity score matching to make it even — so each cohort had over 12,000 patients. And then in the study it ended up showing that patients with rheumatoid arthritis that were on GLP-1 agonists had some decreased chances of having lung cancer and non-Hodgkin lymphoma.

However, when I went a little bit more into it, there was definitely a lot of confounding. We do know that malignancy risk in patients with rheumatoid arthritis comes from disease and disease activity and smoking. So I didn't see — they mentioned smoking adjustment but they didn't mention specifically the amount of smoking, the active smoking — and of course they didn't include disease activity. So definitely a lot of confounding. And then of course a little bit of selection bias, because patients that are on GLP-1 agonists probably have more healthcare access, so probably are more monitored, so things can be a little bit more taken care of than the ones that don't have the access. So for me it was like — yeah, part of the hot topic, but I cannot make a conclusion on this yet.

Well these are very pertinent everyday questions that we need to figure out the answers for if we're going to start using a lot of GLP-1s and incretin therapy in RA as well. And I guess maybe one of the themes of this meeting and others — there's been a lot of Trinetx data, some of it's done well, some of it's done less well. I think this was good, but at the same time it still just generates more questions than answers sometimes.

Anthony: Yeah, these are hypothesis-generating studies so they make us think, but I don't think we can definitively draw a line on them, for the reasons you describe, Adela — they haven't matched for severity of smoking.

I must talk about AI because no talk is complete without some AI. So I've gone for abstract 1348, which is a nice study from Colombia where they looked over nearly 25 years of their electronic health record and they used ChatGPT-4 to see whether they could fish out of 4.5 million total patients how many people had PsA. They gave it 27 features of PsA so that it could pick out how many people met CASPAR criteria. Now, we always had this hype that we could use this to identify people for clinical trials or treatment — if you give the AI the guidelines, it'll tell you how many people they have. So at best only 43% of these PsA patients could be confirmed as meeting CASPAR criteria. It could pick up psoriasis, it could pick up inflammatory arthritis. What it couldn't pick up is disease activity — only 5% of them had enough information for the DAPSA, and only 5% had joint counts. So the message is: if we want to use AI in the future, we better record these things in our EHR, otherwise the data has no use for us later on. The AI can read your notes, but it depends on what's in your notes.

Absolutely. And my final one, going along the same idea of a bit of AI — and I know that Anthony's got one of these, kind of if not quite in your cupboard, you've got it in the back of your office — is Arthritys Diana, the musculoskeletal ultrasound AI-driven robot. I've seen a little bit of this. It's been very interesting how it's all played out in Europe. This was the first independent evaluation, done in Sweden at Linköping. What they did is they looked at 99 patients, 22 joints in wrist and hands in the 28-joint count, and they got an 86% acquisition rate. So it's pretty good. The accuracy is something that's yet to be determined. I think we look forward to seeing data on that. Have you been having fun with yours, Anthony?

Your machine — having great fun. We are collecting data. But I think the performance is improving year on year. We've been to this — you've written on this as well in previous years. So I think it's certainly a help to us in terms of our decision making. It doesn't replace the rheumatologist, but it does help us gain time because we would otherwise be scanning these patients ourselves.

Absolutely. And time is what we all need. Speaking of which, we are at time. It is always a pleasure speaking to the three of you. Thank you very much for joining me. Anthony, Eric, Adela — have a great night. Thank you so much. Very much. Thanks. Thank you.