Loading summary
Transcript38 lines
- [00:03]
Dr. Neal Skolnick
We have another great month of article reviews. I'm Dr. Neal Skolnick, professor of Family and Community Medicine at the Sidney Kimmel Medical College of Thomas Jefferson University. And joining me as always, is my co host.
- [00:18]
John Russell
I'm John Russell. I'm a clinical professor of Family and Community medicine also at Sidney Kimmel Medical College at Thomas Jefferson University. Neal, it's great to be with you tonight.
- [00:29]
Dr. Neal Skolnick
And John, this is another set of great articles. We're going to start with a fascinating article using AI in real world prospective validation and economic evaluation of deep learning based diabetic retinopathy detection from Fundus Photographs. And this is a systematic review and meta analysis of kind of this new world area published in Diabetes Care. Then an article on Orphoglipron, an oral small molecule GLP1 receptor agonist for the treatment of obesity in people with type 2 diabetes. The ATTAIN 2 trial. This was a phase 3 double blind trial published in the Lancet. Then we're going to discuss the removal by the FDA of suicidal ideations for GLP1s. Then an article titled Effectiveness and safety of statins in type 2 diabetes according to baseline cardiovascular risk. And this was a target trial emulation study in the Annals of Internal Medicine. And our final article this month is GLP1 receptor agonists and the risk of Optic nerve or vision threatening events in patients with type 2 diabetes or cardiometabolic disease. A meta analysis of the randomized controlled trials published in Diabetes Care. John, we're going to start with that study on AI real world prospective validation and economic evaluation of deep learning. Deep learning was something I used to think you and I did, but now it refers to AI of deep learning based diabetic retinopathy detection from Fundus Photographs. A systematic review and meta analysis. I know this is something we've touched on before. Deep learning has shown a lot of promise for detecting diabetic retinopathy using fundus photographs. This article sought to assess the feasibility of implementing deep learning for diabetic retinopathy systems using Fundus photographs across different countries, different systems of care. By synthesizing prospective validation and economic evidence. They searched five databases. Studies were prospectively assessed using or looking at diagnostic performance and they looked at economic analysis. They looked in total at 47 studies. This is a lot of information included in this meta analysis. And the pooled performance was highest in detecting vision threatening diabetic retinopathy. The area under the receiver operating curve was 0.974. Just to put that in perspective, above 0.8 is considered good. Above 0.9 is considered really good meaning, has very good sensitivity and specificity. The next best at detecting was any diabetic retinopathy where the area under the curve was Also frankly fantastic. 0.959. 15 studies were included in the economic commentary showing that this was cost effective in high income countries and in middle income countries. The results were mixed. John.
- [04:09]
John Russell
So, Neil, this is something that you and I have been doing in our practice for a while. You know, I think there's a lot of things that you don't get if the only evaluation of someone's retina is just the photograph taken in the primary care office. And they don't always work. So sometimes we'll get kind of notices back that things are inconclusive. But certainly this is infinitely better for our patients who do not find their way to the eye doctors. And there is kind of more and more people who for lots of reasons just are not meeting the guidelines that we recommend to them to get to the eye doctor. So, you know, interestingly enough that this is something that makes most sense in a more resourced country versus a less resourced country. And one would think in a less resourced country that this might, this might make more sense. But so I think it is something that does kind of get more people in that area under the curve for being evaluated. You know, I think as we are using AI for kind of more and more things that we are going to see some of the static things that are looking at images, be it a fundus photo, looking at a pathology slide, looking at an X ray. I don't think things will solely be done by AI. You know, certainly you need some quality assurance being done with some, you know, backup reads and double reads. But, you know, for our colleagues who are ophthalmologists, there's probably more good they can be doing seeing patients live or being in the operating room than necessarily just looking at photos from our office. And so I think that this is an exciting thing and I think we are going to see more of this happening, especially for things that are static.
- [06:08]
Unidentified Medical Expert
So our next article is from the
- [06:10]
John Russell
Lancet and it looked at or fluor
- [06:13]
Unidentified Medical Expert
lipron and a oral small molecule GLP1 receptor agonist for the treatment of obesity in patients with type 2 diabetes. This was a phase 3 double blind, randomized, multicenter placebo controlled trial. So this was a 72 week phase 3 double blind placebo controlled trial that was conducted across 136 sites in 10 different countries. Participants had a BMI of 27 or greater, an A1C of 7 to 10 and they were randomly assigned 1 to
- [06:44]
John Russell
1 to 1 to 2 to oral
- [06:48]
Unidentified Medical Expert
daily or fluorm at 6 milligrams, 12 milligrams, 36 milligrams or placebo. The primary endpoint was the mean percent change in body weight from baseline to week 72, their treatment regimen estimated. Randomly assigned participants regardless of intercurrent events was the primary estimate and with the efficacy estimate considered. Supportive safety was assessed in all patients who were treated with at least one dose of study drug, so this study went from June of 23 to February of 24. There were over 2,800 participants screened. 1600 of the patients were female, which would have been about 46%. They were randomly assigned following a dose escalation phase to receive either 6mg of the or fluorglipron, 12mg 36mg or placebo as an adjunct to lifestyle modification. 1444 patients completed the study. The baseline body weight was 101kg with a BMI of 35 and an A1C
- [08:01]
John Russell
of 8 for the treatment regimen.
- [08:04]
Unidentified Medical Expert
At period, the mean percentage change in the body weight from week 72 was a decrease in 5.1% with 6 milligrams, the estimated treatment difference being a decrease in 2.7, a decrease in 7.0% with 12 milligrams and an estimated treatment difference of a decrease in 4.5 and 9.6% with the 36 milligram dose of or fluorglipron versus overall a 2.5% decrease with placebo. All prescribed weight and Cardiometabolic measures including A1C statistically significantly improved with the Orifluor Glipron. Treatment discontinuation due to adverse events, mostly GI related were higher for orofloor Glipron, so at the at 6 the result 6.1 to 9% decrease versus placebo 4.1%. The most common adverse events with the Orflor Blipron were mild to moderate GI events that would predominantly happen during dose escalation. Ten deaths were reported during the study, six in the treatment group and four with placebo. Investigators deemed all deaths unrelated to the study except for one case in the placebo group and one in the 12 milligram or fluorglipron group. For the case in the OR Flor glipron group, no treatment related association was reported.
- [09:29]
Dr. Neal Skolnick
Neil John I think this is a fantastically exciting medication. So the trial that you just went over Attain 2 was the complementary trial to attain 1. Remember that all medicines that are going to be submitted to the FDA for review for treatment of obesity have to have a trial in people with obesity without diabetes and a separate trial in people with obesity with diabetes. Because in general you lose about a third less weight in all of the trials than the people who are in the trials of people with obesity and diabetes. So attain one which was orphaglipron in people with overweight and obesity without diabetes showed a weight loss of approximately 11% and it had the same or similar decreases improvements in metabolic parameters that all the GLP1s had. Their improvements in waist circumference, systolic blood pressure, triglycerides, non HDL cholesterol levels all improved as we would expect. The other thing in the Tain 2 trial which you just said that's worth note noting, is that the decrease in A1C was quite impressive. A decrease of 1.2% which is in the same range as in the Achieve1 trial, which was orphaglipron in people with early diabetes. In that trial also in the New England Journal last year, mean A1C was 8%. Mean decrease in A1C was 1 point, approximately 1.5% in the people on Orford Glipron. What does this mean? Well, it means we have a new oral medicine that is currently at the FDA right now, as we speak, under expedited review that has particular exciting characteristics. It is a small molecule non peptide GLP1. Why is that important? Well, it's important because its absorption orally is in the rate of 70 to 80%. That's fantastic. Very different than for instance oral semaglutide where the absorption is only about 2 to 3%. Oral semaglutide, remember, just came a couple of months ago to the United States, FDA approved and now has come to market. Also a fantastic oral option for GLP1s. Orphaglipron is different though. Absorption is good and therefore it's easy to use. There are no problems taking it with food or with water. It doesn't have to be taken at a particular time of day. What is also meaningful for this compound worldwide and this is particularly as opposed to the injectable forms of GLP1s, injectables are very subject to temperature variation at every point along the supply chain. They have to be kept at a cold temperature and that is not easy to do. And in many areas of the world a small molecule non peptide doesn't have any of those requirements, which makes distribution which you and I don't think about a lot but it's important if we think about how important of a problem obesity is worldwide and the need for therapy worldwide. So it's going to be easier for distribution worldwide, it's easier for patients to take on vacation with them and for people who prefer not to have an injection. This will, if it gets FDA approved, as these trials suggest it likely will, will be an excellent option for people with obesity and for people with diabetes and obesity. The next piece of information we're going to discuss is a announcement by the FDA on January 13th that they have requested removal of suicidal behavior and ideation warning from GLP1 agonists for obesity. Remember, this was a warning across all medications for obesity when the GLP1s were first approved. And this has not been a warning on the GLP1s for diabetes. Basically this was announced because a very complete FDA evaluation that we'll talk about in a moment did not identify an increased risk of suicidal ideation or behavior with the use of GLP1 medications. And so that they've requested removal of that from the product label. As I said at the time of original approval, that was just labeling across all oral glucose, rather obesity lowering medications. So what did the FDA do to come to this conclusion? Well, they initially reviewed GLP1 clinical trials and found that there wasn't an association between the use of GLP1s for obesity and the occurrence of suicidal ideation. But there were a relatively small number of cases of suicidal ideation observed in the individual trials and that meant there was considerable uncertainty in the risk estimate. So to address that concern, the FDA performed a comprehensive meta analysis of all clinical trials across GLP1s to improve the precision of the risk estimate. There were 91 placebo controlled GLP1 medication trials in the meta analysis, over 100,000 patients. And that meta analysis again did not show any increased risk of suicidal ideation. And in addition, the FDA conducted a retrospective cohort trial using a large administrative claims database comparing GLP1s to SGLT2s in people with type 2 diabetes. That study population, John, was over 2 million people. And I just want to say when I look at how carefully FDA approved medications are examined and then I think of how people tout non FDA approved compounds and things so lightly off of, you know, on TV or in magazines or social media, I think our heads should be examined to even think about those non FDA approved things. So the FDI then looked at a study population over 2 million users and again found no increased risk of intentional self harm on GLP1s compared to SGLT2s, they did not find any increased risk in any subgroup either with diabetes alone or obesity. They then also reviewed, published, observational and pooled studies, same conclusions. Therefore, after that comprehensive review, the recommendation to withdraw this warning. John, your thoughts?
- [17:14]
John Russell
So, Neil, I think for our listeners that this being withdrawn is not going
- [17:21]
Unidentified Medical Expert
to change what they do.
- [17:22]
John Russell
But it's a kind of an interesting concept in that, you know, a medicine comes out.
- [17:29]
Unidentified Medical Expert
And if you think about some of
- [17:31]
John Russell
the medicines during our career, varenicline, which is the most effective medicine for treatment of tobacco addiction and helping people get off tobacco, carried this stigma for a while until the FDA actually pulled this doxycycline. Wow. If you think about Accutane, isotretinoin kind of carries some of these things with it. So if you look, there was a nice review article kind of last year in one of the Lancet journals that actually looked to see, you know, what categories of medicines actually truly kind of carry this over kind of a long period of time. You know, the number one medicine is going to be alprazolam. Number two is Zolpidem. It's going to be some of the medicines that are for anxiety or sleepers are category one. Number two is antidepressants, and there's a wide variety of antidepressants in there. And the third kind of major category is anticonvulsants. So for the context of diabetes, we might have some people who are on something like a gabaket pet or a pregabalin or something for some neuropathy. But, you know, I think, you know, proving causation is really kind of difficult, and people's lives are not necessarily linear. There was a rock and roll star from the early 1960s named Del Shannon. In 1961, he had the number one song in the country called Runaway in the 80s, in the early days of fluoxetine coming out, he was started on fluoxetine for a very severe depression. And several weeks into the course of his therapy, he committed suicide and led to a large lawsuit that was eventually dismissed from his family to the makers of fluoxetine. So, you know, it becomes very easy to kind of take a medicine, and you'd probably rather live in a world where someone's going to say, hey, there's been some reports of this. Let's look at 2 million. Let's look at 2 million people who've gotten this to see if the signal is any louder. And I think with med safety, it's in constant evolution. And I think sometimes when we are hearing some kind of criticisms of medicine and drugs and things like that, it's well how come they didn't know? Well, if something happens 1 in 10,000 patients, you're not going to see it in a study that had 3,000 people in the treatment arm. So I think having post marketing information is what allows us to be as safe as we can. This is something we need to be kind of mindful of, but we should be able to prescribe our GLP1 medicines without worry about suicidal ideation or Our
- [20:43]
Unidentified Medical Expert
next article is from the Annals of Internal Medicine looked at the effectiveness and safety of statins in patients with type 2 diabetes according to baseline cardiovascular risk.
- [20:56]
John Russell
So this was a study that looked
- [20:58]
Unidentified Medical Expert
at participants between 25 and 84 years
- [21:01]
John Russell
of age in the United Kingdom with
- [21:03]
Unidentified Medical Expert
a diagnosis of type 2 diabetes between the years of 2005 and 2016. The patients had no history of CAD, myocardial infarction, stroke, heart failure, failure, myopathy, liver disease, rheumatic heart disease, schizophrenia or cancer. Patients were started on a statin versus non initiation and they estimated the observational analogs of intention to treat effect. Statin initiators were propensity score matched to
- [21:33]
John Russell
non initiators in a 1 to 4 ratio with a Q risk. Three strata of 10 year predicted cardiovascular risk patients were stratified as having a low risk of under 10%, intermediate 10 to 19%, high 20 to 29% and very high greater than 30%.
- [21:55]
Unidentified Medical Expert
They measured the absolute risk differences and risk ratios at 10 years of follow up for all cause mortality and major
- [22:02]
John Russell
cardiovascular death as well as myopathy and liver dysfunction.
- [22:08]
Unidentified Medical Expert
What they found was statin initiation was associated with reduction in all cause mortality
- [22:13]
John Russell
in major cardiovascular disease across all of the strata of risk, the four different categories. In the low risk category the risk
- [22:24]
Unidentified Medical Expert
difference as well as the risk ratio
- [22:26]
John Russell
was a decrease in 0.53 and the risk ratio was 20% lower for all cause mortality and decrease of 0.83 and 0.78 respectively. For major cardiovascular death. A small increase from myopathy was observed in the moderate risk stratum only and there was no associated increased risk for liver dysfunction in any of the categories.
- [22:53]
Dr. Neal Skolnick
Neil John, I'm going to approach my thoughts on this article in two different ways. One is a brief comment on what the article concluded and putting it in context of the ADA's current recommendations and data. That's out there, but then I want to talk for a couple of minutes about target trial emulation studies in general, which is something we're Seeing a lot more of lately. So just realize the risk difference is the absolute risk reduction. Risk ratio we're used to seeing is the relative risk reduction. A risk difference of 0.5 to percent here means that one person out of 200 over 10 years benefited. Now, the benefit was important. It was decreased mortality. And it's always important to look at that rather than just say there's a 20% decrease. So the authors really conveyed that nicely and clearly. This is, it's not a surprise, right, to see that statins work in people with type 2 diabetes. The ADA recommends moderate intensity statin therapy for primary prevention in adults with type 2 diabetes who are over 40 years of age. They recommend high intensity therapy be considered in those with additional risk factors. And this is based on a lot of trials out there. There was a meta analysis of over 17,000 people with type 2 diabetes that showed a approximately 20% reduction in major vascular events for every 40 milligrams per decil liter reduction in LDL cholesterol, which is about what you get with a moderate intensity statin therapy in people with the usual elevated levels of ldl. There have been primary prevention trials, but none of those trials looked at people with very low cardiovascular risk, which is what this trial attempted to do. And the reason no randomized trial looks at that is that the trial would have to last a very, very long time to show a benefit or have to examine a very large number of people. And that's for observational trials. Oops. I said observational trial, John, and they called this a target trial emulation study. Was that a mistake the way that I said it? Well, it wasn't. I said I discussed target trial emulation studies for years. Observational trials, cohort studies, have been what we've used to look at things that are not examined in prospective randomized trials. Our data sets have become more accessible, they've become much larger. And that's led to the emergence of real world evidence as an important form of evidence that we use. A target trial emulation study is a type of observational trial that, that defines ahead of time the inclusion criteria for who's going to be included in the observational trial, how long they're going to be looked at, what parameters are going to be looked at. So it tries to mimic or emulate a randomized trial. But don't let that fancy name fool you. It's still an observational trial. And its main Achilles heel remains the potential for confounding variables to affect the outcome. And when it comes to thinking, are there likely confounding variables. It comes back to all of us using our critical analytics skills to say, what might that be? I think in this trial, frankly, it's easy to think of a confounding variable that invalidates the result. And that confounding variable is that people who are on statins who are on low risk might be either taken care of by doctors that are more aggressive in taking care of every risk factor they have, or might be people who are just more attentive to their own health. And it wouldn't be a surprise that over 10 years they have better outcomes. So I don't disagree with the conclusion that statins are likely beneficial for people at less than the usual 7.5% ASCVD risk or 10 year risk. But I think we should be careful not to confuse target trial emulation studies, which is just another name for observational trials that might have confounding variables, with real prospective studies. For our final article of this month's issue, we're going to look at an article from diabetes care titled GLP1 receptor agonists and risk of Optic nerve or vision threatening events in patients with type 2 diabetes or cardiometabolic Diseases. This was a meta analysis of randomized controlled trials. There's been some conflicting information out there around GLP1 receptor agonists on the risk of ischemic optic neuropathy. So the authors attempted to synthesize current evidence of GLP1 receptor agonists on the risk of optic nerve or vision threatening events from randomized controlled trials in patients with type 2 diabetes or cardiometabolic diseases. They looked at a total of over 83,000 participants from 20 published randomized trials. The primary outcome was a composite of optic nerve or vision threatening serious adverse events, including ischemic optic neuritis, ocular ischemic syndrome, papilledema, blindness, blurred vision, visual impairment, and reduced visual acuity over a mean follow up duration of approximately three years. GLP1 receptor agonist use was not associated with increased risk of primary endpoints. There was an odds ratio 1.2, but it did not reach statistical significance. Prespecified individual adverse events, including ischemic optic neuritis, which had an odds ratio of 1.55 but a confidence interval that ranged from 0.49 to 4.6, and vision loss and disturbance events, an odds ratio near 1, were not significantly associated with GLP1 use.
- [29:44]
Unidentified Medical Expert
John so Neil, kind of an interesting thing, and I think this kind of
- [29:48]
John Russell
ties into what I talked about earlier about some of the risk with regard to suicidal ideation related to a drug. Again, I don't think that this is something that is hitting a lot of the literature that you and I are bumping into month over month. But in kind of prep for talking about this article, I took a dive into some of the vision literature. And certainly this is something that there are papers by the American Academy of Optometry and the American Academy of Ophthalmology, and there are kind of signals that are being kind of raised in some of these professional organizations about some of these medications.
- [30:35]
Unidentified Medical Expert
And I think that this is also
- [30:37]
John Russell
a great example of some of the checks and balances, right? So if you look at that ischemic neuropathy that can happen, it happens, you know, pretty rarely, about one in every 10,000 patients. So not something that someone's going to see very often in kind of a smaller study. But I think this meta analysis is really reassuring and I think ultimately, you know, without kind of going through every different eye condition, at the end of the day, they could say with these 83,000 patients that they found that no one was at increased risk of anything that lost their vision. And, and in fact, you know, overall, kind of depending on where we are in the course of someone's diabetes, better glycemic control, which is going to be attached to all the GLP1s, is going to lead long term to better retinal health, better microvascular disease control. You might not see it kind of later after someone has already developed that and even they kind of cut out that there really wasn't. It really didn't make a difference if someone entered into one of these studies with pre existing retinopathy or not.
- [32:07]
Unidentified Medical Expert
So I think that this should be
- [32:08]
John Russell
reassuring that the signal that we're hearing, at least at this point, is really not associated with it. And but they cannot completely rule out things that happen in a very low incidence. So again, if this meta analysis had 80,000 people in it and there's something that happens, one in a hundred thousand people, the study was not put together enough to pick up kind of all of those events that are more rare than the end that was in the study.
- [32:38]
Dr. Neal Skolnick
John before we conclude, I just want to remind all of our listeners to listen to a new journal podcast that is now on our feed. And that podcast is of the new American Diabetes Association's journal, Diabetes, Obesity and Cardiometabolic Care. Really some fantastic information. And this month that podcast is going over the new ADA's Obesity Association's standards of care for overweight and obesity. Really fantastic podcast. Take a listen. For more information and links to the articles that we discussed in this issue, just go to diabetesjournal.org until next month,
- [33:30]
John Russell
keep listening and keep keep learning.