← Endocrine Feedback Loop
Loading summary
Transcript27 lines
- [00:01]
A
This is endocrine feedback loop. I am your host Chase Hendrickson and welcome you to this Journal Club Podcast series brought to you by the Endocrine Society. Thanks for joining us as we explore an important article recently published in one of the Society's clinical journals. Welcome back to the Endocrine Feedback Loop podcast for our 64th episode today we return to the JCNN to review a recent paper that attempts to elucidate the connection between subclinical thyroid disease and cardiovascular disease. Any of you listeners who see patients with thyroid conditions likely frequently encounter people with mild thyroid abnormalities and need to figure out if those abnormalities might have a clinical impact. So given the many unanswered questions in this area, we thought this article well worth discussing. Not only do these authors use an observational study design, but their work is a pooling of multiple observational investigations, necessitating a careful thinking through of that methodology. I host the Endocrine Feedback Loop and work at the Vanderbilt University Medical center as a general endocrinologist and Medical Director. Two excellent thyroidologists join me for this episode of the podcast. Anupam Kotwal returns as our regular contributor for the month, hailing from the University of Nebraska Medical Center. There he focuses his clinical care and research on endocrine oncology, especially thyroid cancer. He is the medical director for their thyroid tumor program and has a research emphasis on immune biomarkers for thyroid cancer. Today's guest expert is Jenna Mammon from Johns Hopkins University. She focuses her practice on thyroid disorders with extensive publications in the field. Her research looks at the effect of aging on thyroid disorders as well as pituitary and adrenal conditions. So, as usual, the perfect pair of endocrinologists joins me today. As is also always the case, everything we say will be our opinions only and not those of our respective institutions or of the Endocrine Society. For today's episode, we look at associations between subclinical thyroid dysfunction and cardiovascular risk factors according to Age and Sex, which the Journal of Clinical Endocrinology metabolism published in May 2025. Oliver Bartella and Manuel Blum at the University of Bern represented the Thyroid Studies Collaboration as the first authors for this manuscript. I will now hand the conversation over to Anupam, who will go through the author's introduction and get Jenna to go in depth on a few key concepts.
- [02:34]
B
Thank you, Chase. I will start with some discussion about the introduction and the background for the study. Subclinical thyroid dysfunction is defined as a thyroid stimulating hormone or TSH level that is outside the normal reference range, but thyroid hormone levels are within the respective reference range. It has been shown that subclinical hypothyroidism as well as subclinical hyperthyroidism are linked with a higher risk of cardiovascular disease. However, whether subclinical thyroid dysfunction is actually associated with the risk factors for cardiovascular disease is unclear. Some studies have shown that subclinical hypothyroidism is associated with a higher risk of hypertension, obesity and dyslipidemia, but this data has not been corroborated in other population based data. Also, treatment of subclinical hypothyroidism has not been shown to significantly impact lipid panel or dyslipidemia. On the other hand, there is risk of subclinical hyperthyroidism also with specific factors associated with cardiovascular disease. So this study is evaluating the link between subclinical thyroid dysfunction and cardiovascular risk factors including blood pressure, lipids and high sensitivity CRP or C reactive protein, as well as smoking status according to age, sex and TSH levels. So here I would like to ask Jenna to briefly summarize these available findings that we have in the literature and also what our current understanding is of the link between subclinical thyroid disease and cardiovascular disease.
- [04:19]
C
Thanks Anupam and thanks Chase for having me. So this is a big topic and obviously of great interest because we have a huge number of patients and especially older patients that come to us with mildly elevated TSH levels and normal free T4 levels and there's always a question about whether or not these patients should be treated. And the literature is quite mixed on this point. The most recent meta analysis was done in 2018 looking at cardiovascular endpoints and found that overall there is an increased risk of fatal and non fatal cardiovascular outcomes in patients with high risk of cardiovascular disease and not a significant risk in lower risk patients. But also this risk mostly accrues to younger patients and not to older patients. And mortality overall all cause mortality was not associated with subclinical hypothyroidism in that setting. Now we're going to talk about this a lot I think in this session today, but one of the problems overall with all of the studies that have been done to date and with this study too, is that by and large these are associational studies that are trying to establish a link between an outcome that we care about and an exposure, but aren't actually manipulating the exposure or testing whether changing the exposure to subclinical hypothyroidism affects the outcome and really the only way to show whether or not there's a causal association is by doing a randomized control trial where you take some people who have have the disease and give them a treatment for it and other people who don't and see whether or not you can affect the outcome that way. So there was a big study that was set up not that long ago called the Trust Trial, that was actually going to try and ask this exact question. Could treating subclinical hypothyroidism impact cardiovascular disease? And could we improve patients cardiovascular outcomes by treating subclinical hypothyroidism? And it was a big multinational study. The problem is it never got powered to actually answer this question. And the reason it never got powered to answer this question was because a lot of people with subclinical hypothyroidism actually don't persist in having subclinical hypothyroidism when you test them multiple times. And this brings up a second point with associational studies, which is what is the cause and effect is actually very tricky. Or because you don't actually know what the relationship is between the two things that are associated. In older adults with mildly elevated TSHs, there's actually a growing literature that suggests that some of those elevated TSHs are not actually primary thyroid disease related as we get older and have increased systemic inflammation and other kinds of things that maybe our thyroid axis is actually reacting to that in an adaptive way. And that elevated TSHs and lower T3s might be a sort of mild non thyroidal illness syndrome pattern that's trying to be counter regulatory and homeostatic to that increased inflammatory disease. And if that's the case, then an association with poor outcomes actually might be the reverse cause rather than the direct cause. Because of course, people who are at higher risk of cardiovascular vascular disease are going to be exactly those people in whom the thyroid axis is going to be trying to downregulate our catabolic effects and trying to decrease the stress on the system by having that non thyroidal illness pattern. And so reverse causality is another problem in this field. And so it's a very tricky question to answer. And that's why I think we're stuck with still trying to dig down into these questions of these secondary endpoints and seeing if we can find a mechanistic reason why there might be an effect of subclinical thyroid disease. If there is an effect on blood pressure, if there is an effect on lipids, then maybe that would give us a reason to act even if we can't demonstrate a direct effect on the cardiovascular disease.
- [09:10]
B
Thank you for very nicely highlighting the challenges in evaluating these clinically important questions. So I will again reiterate what the study is aiming to do. It is a pooled analysis aiming to evaluate the associations between subclinical thyroid dysfunction, both hyper and hypothyroidism, with cardiovascular risk factors, including blood pressure, lipids, high sensitivity, CRP, and smoking status, according to the participants, age, sex and TSH levels. And with that, I will hand it off to Chase to discuss the methodology of this study.
- [09:48]
A
Thank you. Anupam and Jenna, you guys nicely outlined the questions that we have here and why we have to look at this observational data. As difficult as it can be to interpret. We'll start thinking about that with the study design. And I would argue that the way the authors construct the this research is that it is a cross sectional study. The term in the paper that may get thrown around a little bit is cohort. But one thing that I would highlight and why I think this really should be viewed as a cross sectional study, is that cross sectional studies, all of the data is collected at a single point in time. Jenna nicely highlighted the idea of exposures and outcomes. We're always thinking about those. And a key with cross sectional studies is that you're getting that data all at a single point in time. And you may have a hypothesis that we think this is the exposure and we think this is what may be driving the outcomes of interest. And you've got that, that causal effect outlined there at least proposed. But one of the challenges is, is because you're collecting all that data in time at the same point in time, you often don't know that one predates the other. In a cohort study, you would be able to demonstrate that we have these exposures. Nobody has yet developed an outcome of interest, but we can follow people over time and then when they develop the outcome, we can see if one group has that more than another. Cross sectionals, you're usually very limited on that. We'll get into this. There is a little bit of data over time for this group, but it's actually exposure data. It's not the outcome. So I think that's going to be important to keep in mind. So as far as this study goes, Anapom's mentioned it already. This is actually a pooled analysis. So these authors were able to tap into a database of multiple studies that had individual participant data. So they were able to pool all that together into a single group. Again, the authors describe the studies that they're pulling from as cohort studies. But again, I think this is really the way they analyze it makes a lot more sense to think about it as a cross section sectional analysis. So they only looked at studies that included TFT. So you had to have a TSH and a free T4 and those cardiovascular risk factors that Anupam mentioned already. So the blood pressure, lipids, that high sensitivity crp, the smoking status. And then you also had to have information on age and sex. And the, one of the important things to consider here is who they excluded. So Jenna's going to help us think through this in just a second, but I'll, I'll mention that they did not want to look at anybody who would be treated for any of these conditions. So if you were on an antihypertensive, if you were on any sort of lipid lowering medication or on a thyroid medication, and that also included if they didn't know, if the authors couldn't determine what medications you were on, then you were also excluded since you might have been treated. So, Jenna, I'm interested in your thoughts on this, this approach. The authors are really interested in excluding the effect of medications out of worry that, well, if we treat something, then maybe it's going to effectively treat it and we can't really see whether something is there or not. So they said we're going to get rid of that. But one of the concerns in my mind is that you're necessarily getting rid of a, of your high risk people. And so it may be a lot harder to find an association. So obvious trade offs and having to make that decision. But, but what are your thoughts on the author's decision here, the way they design this investigation?
- [12:53]
C
This is a very challenging question, right, because if somebody's treated with an antihypertensive medication, then their blood pressure is going to be lower. If your outcome is the actual blood pressure, then they're not going to have a high blood pressure because they're treated. If your outcome were a diagnosis of hypertension, you could have a combined definition that included being on an antihypertensive medication or having an elevated blood pressure that was untreated. And so there are ways to include patients who develop a condition and get treated for it in your outcome definition. And that's not what their outcome definition is here. And so you're right, they're excluding everybody who has a definition of hypertension which may or may not have been related to their untreated thyroid disease, where their blood pressure or their Cholesterol got treated but their thyroid disease didn't. And what was the reason that their primary care doctor made the decisions to put them on a antihypertensive or a lipid medication and not a thyroid medication. Or we can't know that that's the decisions that the primary care doctor made. Similarly, we don't know why some people got treated for their thyroid disease and other people didn't get treated for their thyroid disease. And so there's also selection bias even in the exposure versus the unexposed group and the exclusion of people based on being treated for their thyroid disease. Because maybe the people who have thyroid disease and end up on thyroid hormone are different from the people who have thyroid disease and don't up on thyroid disease in particular, with regard to their cardiovascular risk factors. Right. If I see a patient in clinic and they have mild thyroid disease and they also have a high risk of mi, maybe I would be more likely to treat them because I'm worried about their cardiovascular risk factors and I want to treat them for their thyroid disease or especially if they have hyperthyroidism, I might know about the AFIB and think, oh, I really need to treat that person for their mild hyperthyroidism, or if they have mild subclinical hypo, I might be worried about over treating them and not treat them. And so under treat people with cardiovascular risk factors for hypothyroidism. I don't know what the answer is. That's not answered in this. But there's certainly a risk of bias in the fact that people are being chosen for treatment or not for treatment by their primary care doctors. And then the people who are on treatment have to be excluded from the study. And so because there are people on treatment in the study and they are, you know, in these cohorts, they have people who are on treatment and they know about what percentage of them are on lipid lowering medications and what percentage of them are on antihypertensives. They actually probably could give us some information about how different the excluded patients were from the included patients. And sometimes that's a way for studies like this to reassure us that there's not a selection bias in terms of the cohort, that the excluded patients are similar to the included patients. And that data is not actually provided here, but would have been the kind of thing that would have been helpful in terms of the sensitivity analysis to really understand that we're dealing with a relatively representative population and it's sort of random based on individual physicians making these decisions. Some people decide to treat other people. Don't we get kind of a good sample, Especially given how large this population is and the fact that it goes over multiple continents, we might anticipate that it's so many different individuals making decisions that it comes out in the wash. And so it's possible that there's no difference, but we don't know that.
- [16:50]
A
And Jenna has nicely highlighted how we think about biases here. And typically the usual way of thinking about a bias is trying to decide whether somebody get treated or not treated because of that bias. But as she's given us some good examples of, you could also think about that exact same phenomena with who got excluded from this study. Okay, now going back to that terminology of exposure and outcome. Well, think about the exposure. And the authors put these individuals into three groups based on exposure. They describe them as subgroups, which I think is one way of thinking about them. But I think, again, it's just degree of exposure. So they have individuals who have subclinical hypothyroidism, who have euthyroidism or have subclinical hyperthyroidism. And some additional analyses that they did based on that construct is they later went on and did some sensitivity analyses just to say, well, we've grouped people in this. Were there any decisions that we made that maybe impacted the outcomes that we saw? So they also planned sensitivity analyses by looking at individuals who had persistent subclinical hyper or hypothyroidism. Only a few people had that. But they wanted to look and see if they would have come to a different conclusion if they had exclusively looked at those individuals. They also looked at people who had more marked abnormality. So the definitions for subclinical hyper and hypo were the standard ones that we use. But when they looked at the more marked abnormalities for subclinical hyperthyroidism, that included individuals who had a TSH of less than 0.1, and then for subclinical hypothyroidism, individuals who had a TSH of 10 to 20. Another way of thinking about that would be, well, TSH is where often we would think about treating, but in this situation, again, we're not treated. Anupam's mentioned already some of the stratification that was done here. So there was this concern about confounders that we've talked about already. So one way of dealing with confounders is to stratify, is to put those people into different categories so that you can eliminate that effect. So they did that. For both sex and age. And of note, with age, that was whether you were less than 70 or greater than 70 years of age.
- [18:45]
C
Anytime you stratify and cut things down, you lose power. And that's one of the issues, too, with doing the dose response effect, which is absolutely one of the things we want a study like this to do, because if there's something mechanistic, we would certainly expect there to be a dose response. Right. A dose response is one of the ways we really sort of validate for ourselves that there's a causal process going on. It's one of the epidemiologic pillars of demonstrating a causal relationship as a dose response. So by stratifying by the more severe and the less severe and seeing if the effect is related to the degree of disease, we can get a good sense of it. But often there aren't really that many individuals who have the more severe your disease category. And so sometimes we lose a lot of power doing those analyses. So, you know, it's definitely something that we want to do, but we do have to sometimes take it with a grain of salt if the results turn out to be negative.
- [19:50]
A
Finally, we'll think about the outcomes. And we've mentioned this before, but I think this is worth highlighting, is that the outcomes here are cardiovascular risk factors. So they're not actually cardiovascular outcomes as we would traditionally think about them, but risk factors for cardiovascular outcomes. So, Jenna, let's go back to you one more time here in the methods. Help us think through that. So this isn't a study looking for who had myocardial infarction, who had death related to a cardiac event, but this is cardiovascular risk factor. So walk us through why the authors might have included that kind of the pros and cons of that approach.
- [20:26]
C
As we said at the beginning, what we really care about, obviously, is keeping our patients healthy and preventing cardiovascular disease. And there are a lot of good studies, Framingham study and that kind of thing, that show that the risk factors and modifying the risk factors like your cholesterol and your blood pressure can improve those outcomes. And so if thyroid disease is impacting on your cholesterol, then it would be a good reason to be more aggressive with your thyroid, because that would be helpful. We talked a little bit at the beginning about the Moon meta analysis that showed that in high risk people there was more of an effect. It may be that if your cholesterol's already basically within goal, lowering your cholesterol by 10 or 15 points with your thyroid isn't gonna do anything. But if your cholesterol is really high, lowering your cholesterol even a little bit more and helping the statin out a little bit more would be helpful. On the other hand, it's also possible that, you know, you're not going to get anybody off of their statin by lowering the cholesterol 10 points using the thyroid hormone. And so we do want to be cautious about, like, what is the effect size and how much does this really matter? And sometimes with these massive studies, we can measure a statistically significant effect, but is it clinically significant? And so there's two things. When we measure these secondary outcomes, these intermediate outcomes, it's both, is it statistically significant? Can we measure an effect and then is it a clinically significant effect? If we do this, is it really going to impact on our patients and at what cost?
- [22:15]
A
And those last points will be particularly important to keep in mind as we move into the results. I now will do that. I'll hand it back off to Anupam, who's going to walk us through the results as the authors present them.
- [22:25]
B
The study included about 69,000 participants from 16 cohorts in the Thyroid Studies collaboration, spanning North America, countries in Europe as well as Australia. On this pooled analysis and this, as shown in the Table 1 of the study, the mean age of the participants was 62 years, ranging from 18 to 103 years. So all adults, 55%. So about half of the overall cohort were women and 25% were current smokers. Amongst this cohort, 5% had subclinical hyperthyroidism and 5.4% had subclinical hypothyroidism. So fairly similar prevalence there. Mocked subclinical thyroid dysfunction was less than 1%, so 0.9% had marked subclinical hyperthyroidism, which is a TSH less than 0.1 international units per liter, and then 0.6% had marked subclinical hypothyroidism, which is a TSH between 10 to 20. Again, they excluded participants with a TSH above 20. In this cohort, 4.5% of participants were taking a medication to treat a thyroid disorder, either thyroid hormone replacement or antithyroid medication. About one third took antihypertensives and 12% took a lipid loading medication. Again, the participants who were taking these medications were then excluded from those specific analyses. As we discussed. Overall, this is, you know, in terms of demographic composition, this would be similar to what we would expect from participants having subclinical thyroid dysfunction. Again, most of that is usually not marked, but there are some ways that they have excluded specific participants from the results. So now I will go into the specific associations here. We'll start with blood pressure. So the authors are not reporting on either the mean blood pressure or the lipid values here because they were not adjusted for age and smoking status in the studies. In the adjusted analyses, they found that compared to youth thyroid participants, women with subclinical hyperthyroidism had a lower diastolic blood pressure and men with subclinical hyperthyroidism had lower systolic blood pressure. But again, these values are of very low magnitude, 1 to 3 millimeter mercury. So something to consider and keep in mind as we go into discussion later, to me, I think more impactful is their analysis, which restricted to marked thyroid dysfunction. So again, TSH less than 0.1 or 10 or more. And in that group, group, it was seen that although men had similar blood pressure than euthyroid participants, so men with marked thyroid dysfunction did not have any significant blood pressure difference than euthyroid. The women with marked subclinical hyperthyroidism had a higher systolic blood pressure by 3 millimeter mercury, while those with subclinical hypothyroidism had a higher diastolic blood pressure by 2.5 millimeter mercury. So I'll take this opportunity now to ask Jenna for her thoughts on these findings based on what we understand about the effects of thyroid function on blood pressure and how that much that is clinically significant.
- [25:48]
C
Thanks, Annapon. So there has been a meta analysis in women on blood pressure studies that showed that there's an increase in the diagnosis of hypertension among women with hyperthyroidism and some clinical hyperthyroidism. So, you know, you would expect there to be an effect. I think that this gets to the small effect and how hard it was for them to find is, you know, a reflection of what we were talking about before. 30% of the cohort is on blood pressure medications, and so was excluded. So we're just talking about people who are not being treated for hypertension and don't have a diagnosis of hypertension. And what is the effect of thyroid disease on people who otherwise don't meet criteria? And that's kind of a different question than does it cause hypertension or what's the overall impact? There's probably a healthy selection here. And even so, in the most severely impacted patients, we are seeing a small increase in the systolic blood pressure. So it's consistent, I think, with the literature that suggests that there's a, an impact and there's, you know, plenty of mechanistic reasons to think that hyperthyroidism would cause a mild increase in, in hypertension. Whether it's clinically significant or not, I don't know, and certainly it's not. The only reason that we would give for treating people with marked hyperthyroidism would be, you know, a one or two point difference in the blood pressure. It's not enough to raise a lot of eyebrows here, but it's certainly consistent overall with the literature why there's a sex difference. Again, you know, it's hard to know, you know, are men more likely to be treated for hypertension than women? And so the, the people who are vulnerable are weeded out. You know, they don't give us enough information about the excluded participants for us to really be able to know whether the it's selection bias that's causing some of those sex differences.
- [27:59]
B
Moving on to the lipids and inflammation, just for context, the data reported in the study are in millimole per liter. If we use milligram per deciliter, which use in North America fairly commonly, it would just be about multiplying the millimole per liter value by 38.67. So 1 millimole per liter of cholesterol is equivalent to about 38 milligram per deciliter of cholesterol, either LDL, HDL or total. With that in mind, in the adjusted analysis in this study, both men and women had comparable lipid values across all thyroid categories. Now, women with subclinical hyperthyroidism had a lower LDL cholesterol compared to EU thyroid participants. But in their sensitivity analyses, which they looked at persistent subclinical thyroid dysfunction, this difference was not present anymore. And then their second analysis again is looking at marked thyroid dysfunction. As we discussed in that group, it was seen that compared to EU thyroid participants, both men and women with marked subclinical hyperthyroidism had lower LDL by 0.48 millimole per liter or 18.6 milligram per deciliter in women and slightly less 0.4 millimole per liter or 15 milligram per deciliter in men. So some difference in LDL cholesterol in those with marked subclinical hyperthyroidism, showing lower LDL cholesterol than euthyroid participants. But again, small value there on their subgroup analysis by age amongst participants with age less than 70, again those that were male and had subclinical hyperthyroidism showed a lower blood pressure actually compared to euthyroid participants, while the lipid values and high sensitivity CRP were similar across TSH categories. And when they looked at participants with age of 70 or more, they did not find any clinically meaningful differences in the blood pressure, lipid values or hscrp, either clinical or statistically significant differences there when they compared to euthyroid participants and there was not even any difference even when they used a TSH cutoff of 7.5 or more. So we know that in older participants the TSH cutoff or upper range changes. So even when they used a cutoff of seven and a half to define subclinical thyroid dysfunction, they did not find any difference in these risk factors compared to euthyroid participants. So with that I will hand it back to Chase to go over some discussion.
- [30:50]
A
The authors take all of these findings that Anupam just walked us through and they summarize it. I'll quote them here by saying cardiovascular risk factors were similar between the subclinical thyroid disease and euthyroid group. Differences were of only marginal clinical relevance and irrespective of age and sex. The authors then go on to unpack that just a little bit. So starting with the lipids, they do point out that as Anapalm has told us, that marked subclinical hyperthyroidism did have lower total NLDL cholesterol and that the subclinical hypothyroidism did not have unfavorable lipids and potentially due to excluding those on treatment. And we've talked about that several times already, they move then move on to blood pressure, talk about that, pointing out that women with marked subclinical hyperthyroidism did have higher systolic blood pressure values. And after that they then compare, as is traditional to the medical literature. So other studies that are similar that have been published in the past and try to point out similarities and differences. Jenna mentioned in the introduction that the literature is a bit all over the place, so they try to figure out how that aligns with what they have found here. And they mentioned that other studies that showed that subclinical thyroid disease was associated with an increased cardiovascular risk, but that this study showed only these marginal differences. They also point out that with those other studies showing that subclinical thyroid disease was associated with heart disease, cardiovascular events and even mortality, that it could be that the explanation is that whatever thyroid is doing, high levels, low levels, it's an independent risk factor. In other words, it's not being mediated by these traditional cardiovascular risk factors, the ones that they looked at here. So, Jenna, question for you on this. I guess a couple of questions related. So interested in how you would compare this to the medical literature, either the literature that the authors cite or other studies that are out there. And then as a follow up question to that, interested in your thoughts on the author's mechanistic explanation here? Does it, do we think that's what it fits, that thyroid's doing whatever it's doing through some other mechanism, potentially directly.
- [33:01]
C
So I've talked a fair amount already about my concerns about the selection bias. And, you know, I think it's really hard to study thyroid disease because some people are treated. A lot of people are treated. I mean, we know in the US Thyroid hormone is the fourth most commonly prescribed medication. There's a lot of contamination in that sense of the pool of people who have subclinical thyroid disease. And so it's really hard for us, I think, to do a clean study and know whether the people that we're looking at really represent the risk pool. And again, here, if you're excluding people who are already on statins, you're not really telling us whether or not the subclinical hypothyroidism that they have is affecting their lipids, because some people were treated for lipids. There is a paper actually, Dr. Kotwell did very nicely about five years ago that looked at the literature on treatment, the randomized controlled trials with thyroid hormone treatment and lipids as an outcome specifically. And again, it's not cardiovascular disease as an outcome, but it is higher quality Attempts to understand if you take people with subclinical hypothyroidism and you compare treatment and no treatment in similar patients, do you impact on the cholesterol? And that paper showed pretty clearly that the answer is yes in randomized controlled trials. I think that There were about 23 randomized control trials for lipids and over 100 papers overall that are observational studies on lipids, that overall, the answer is yes. Subclinical hypothyroidism is associated with higher levels of lipids. And you can bring that down, those lipids down if you treat, treat. But the size of that effect is about 10 points, which is about the same size that's being seen in this study. And so, yes, there's a statistically significant effect on lipids in that study. But is there a clinically significant effect on lipids? Are we going to avoid treating people with statins if we treat them with the thyroid Hormone, the answer is probably no. And so if you have a study where 12% of of the population is on statins and that's the people who had the high cholesterol, then that's the impact that the thyroid had on your lipids. Of course that's going to get rid of it and you're not going to see the impact in the people who are left necessarily. So I'm not sanguine that this study gives us better information than the RTCs already do, or that it gives us leave to say that thyroid hormone would be working through a different mechanism. We know that thyroid hormone is a major regulator of lipid metabolism. You know, it, it manages beta oxidation and it's in charge of acyl carnitines and triglycerides. And so I don't see a reason to think that thyroid hormone doesn't manage lipids. I think that the fact that we didn't see a strong signal in subclinical hypothyroidism really points to a potential study design weakness here. What we expect to see in the extreme subclinical hyperthyroidism group actually supports that there's a study design problem in the subclinical hypo, because actually, that's the population that I would think would be least contaminated by selection bias, because those are people probably with graves disease or toxic nodules that haven't yet been treated some other thing. And that if you get to that point, your doctor's not paying attention, and so those people are probably less likely to be treated for anything. So my guess is that if you actually have severe subclinical hyperthyroidism, you're probably not seeing your doctor often. So, you know, the fact that they find it in the population that's probably underutilizing healthcare and not in other populations makes me hesitant to really change practice based on these results. But on the other hand, you know, our practice was not really that strongly driven to treat subclinical hypothyroidism in older adults, based on the literature. Right. I mean, we talked about that at the beginning. Older adults in the meta analysis by Moon didn't show a particularly strong effect. The effect, if any, is really in younger patients. The association is mostly with people who have high risk. And it's not clear to me that that's not a reverse causality effect. So I don't treat patients with elevated TSH based on their cardiovascular risk factors. That's not a driving factor for me in my decision making necessarily.
- [37:59]
A
And at the end we'll come back to thinking about our clinical practice and if this moves us one direction or another or just confirms what we're doing right now. First though, we'll go through what the authors describe as their limitations and I'll list those. First of all, they point out that in many of the studies they include there was an absence of medication data and so they necessarily had to exclude patients when that occurred. They do point out that they don't have any information on HRT and then readily acknowledging the impact that that could have and that they don't know that data for these patients, these individuals. Next, they point out that in these studies there was not any information on acute illness or recent use of iodinated contrast and while unlikely to be a frequent issue, it might have occurred on occasion and they could not detect that. And then finally, as we've talked about already, they point out that for the vast majority of these individuals, thyroid function was assessed at only a single point in time. They didn't see a big difference in handful of individuals where they had multiple points of time, but that was not the case. They didn't have that information for most of these people. Finally, the authors conclude by stating subclinical thyroid disease and EU thyroid have similar cardiovascular risk factors and differences are arguably too small to explain the increased cardiovascular risk in subclinical thyroid disease observed in previous studies. So as advertised, in a minute we'll get here to whether this should change our clinical practice. But but first let's just wrap up by thinking about the quality of this report overall. Anupam, what are your thoughts on that?
- [39:30]
B
There are definite strengths of this study. I would say the, the pooled analysis from multiple studies from multiple sites and different continents, I think is, is a strength. Specifically their stratified analysis. I like that they stratified for not only the participant sex but also age and using that age seven cut off and specifically evaluating the TSH of seven and a half to kind of define subclinical hypothyroidism in that group. Those are strengths there because those are clinically meaningful. We in some respect when we see the patient in clinic and are having that decision making for treating subclinical thyroid dysfunction. Of course we've discussed the limitations so I would say it's kind of moderate to good quality in that setting with clear and the authors do highlight the limitations and what would be the next steps that there. But I think they are attempting to answer some clinically meaningful questions and some of it is confirming what we already know. Which is also great. But that's, I guess, how I would put this currently pooled analysis where it fits in the literature.
- [40:40]
A
Jenna, same question for you thoughts on the quality of this report. I agree.
- [40:44]
C
You know, when, whenever we get a lot of patients in a big study study, it it's very helpful to be able to look at multiple studies together. And doing meta analyses in multiple ways is helpful because every individual study has limitations. And so to be able to combine studies across those limitations gives us insight into what the overall picture is that we can't answer in individual studies. But I think it's a problem with the thyroid field in general that we just are struggling with how much people are treated and how hard it is to really get a clean sample and a clean question. And none of the cohort studies are designed yet to really answer that question. You know, we don't have people who are recruited at the time of their diagnosis of subclinical hypothyroidism and then observed untreated off of treatment with or, or, you know, that's not who these are. These are just people who are recruited and then you just see what happens to them. And so their treatment is at the hands of their primary care doctors. And that treatment is so variable and at such a high rate. I mean, one of the interesting things actually, overall the rate of subclinical hypothyroidism in this study is 5.4%. But in NHANES in the United States or in the Colorado symptom survey from 20 years ago was actually more like 10%. If you take a random sample of the US you should see about 10% of people with subclinical hypothyroidism. So actually, overall, it's a little low in this study, which suggests that there's a lot of treated hypothyroidism in the study. And maybe these are, you know, who participates in a study who says, yes, I'll volunteer. Right. There's that factor too. So are these people who are more likely to be involved and more likely to get their blood pressure and their cholesterol treated. So, you know, that would all mitigate against making an observation. And so I think it's hard to take a negative result from this kind of a study and say that if you looked in the general population in a real world data set where things were unselected, that you would necessarily find the same thing.
- [43:01]
A
Audifon, final question for you. You've written a lot about this situation. You of course see lots of patients in this situation is the results of this study. They're gonna change what you do right now in clinical practice, confirm what you do. Walk us through how you're gonna think about that.
- [43:15]
B
So I do think their results do confirm and kind of just strengthen how we approach treating subclinical thyroid dysfunction. So their results showing that in participants of age of 70 or more, you know, even if their TSH is above seven and a half, it may not be meaningful or impactful to give them thyroid hormone replacement unless they have symptoms and things like that, because it may not have a significant change in any of their cardiovascular risk factors. So I think that kind of confirms that in that age group it may be more risky than beneficial of treatment. Mildly elevated TSH levels with normal thyroid hormone T4 levels. So it does confirm that and strengthen our support of that practice. And then the other thing is, you know, what is the difference in cholesterol and blood pressure that comes with this? So again, this study, as well as prior studies that we discussed, there's a very minimal change in the LDL cholesterol and total cholesterol when we treat. Treat subclinical hypothyroidism. Now, if a patient is sitting in front of us in the clinic, the way I would look at it, if they had severe enough dyslipidemia or hypertension, and if they're already on medication or they need treatment for that, we evaluate it. And when we treat that, if we've deemed that they don't have severe enough dyslipidemia or hypertension to require a specific medication, then we can say, based on this study and other studies, studies, it may not be meaningful to treat their subclinical thyroid dysfunction to improve the cholesterol or blood pressure, because even if it improves those parameters, there will be only a slight difference. Now, who will need the actual lipid lowering medication or antihypertensive medication. Maybe there's a bunch of factors, including possibly subclinical thyroid dysfunction, but that group was not looked at in this study. So. So that's how I look at it. I think it strengthens our clinical practice of not treating subclinical hypothyroidism, especially in participants above age 70, unless they are symptomatic, because it can be more risky in terms of having TSH suppression and that actually having cardiovascular risk. When we look at subclinical hyperthyroidism that we usually do want to treat because of its cardiovascular and bone adverse effects. And in that case, the age group above 60 or 70 is actually more prone to have those adverse effects. So there we actually have a lower threshold for treating subclinical hyperthyroidism, but that is not to improve or, you know, change their blood pressure or cholesterol, but really to decrease the risk of atrial fibrillation and bone loss. So that's how I would think of these data supporting our clinical practice.
- [46:11]
A
Jenna, you alluded to it before, but would you take a different approach on that or do you have a similar conclusion on how this should change our clinical practice?
- [46:19]
C
Yeah, no, I totally agree. Anupam, you summarized it very nicely.
- [46:23]
B
Right.
- [46:23]
C
For hyperthyroidism, our primary concern is the bone and the arrhythmias and certainly a little bit of hypertension, which we're finding in this study, would exacerbate any arrhythmic risks. And so this strengthens and supports our concerns about subclinical hyperthyroidism and the, the reasons to do that. Stroke is another risk factor. With hyperthyroidism, we have some concern about dementia and hypertension, so another risk factor for dementia. So all of that sort of goes towards the idea that subclinical hyperthyroidism, especially in older adults, is probably not so good and that we should probably be more aggressive, maybe even than we are in being sure we treat subclinical hyperthyroidism. And on the subclinical hypothyroidism side, not only is there no evidence that it really needs to be aggressively treated, especially in older adults, these are still associational data and there's still the question of which way the association goes and whether it's really causal and whether, as Anupam was saying, if there's no evidence that having an elevated TSH in an 80 year old is associated with bad cardiovascular outcomes or higher lipids or higher blood pressure, then maybe it actually doesn't mean that there's really a causal relationship there at all. And that what we've seen for us, you know, in, in some of the associational studies is just the fact that it's not subclinical hypothyroidism in some of those patients, it's the sicker patients have these HPT axis changes that are adaptive. And so if you study sick people, you're going to find an association that's not present when you just study healthier people.
- [48:21]
A
And with that, I would like to thank Anupam Kotwal and Jenna Mammon for joining me for this month's edition of Endocrine Feedback Loop. I hope that you all learned as much as I did and that you will join us again next month. And now you're in the loop. This has been Endocrine Feedback Loop. Endocrine Feedback Loop is brought to you by the Endocrine Society with Production oversight by Brandy Brown and Andrew Harmon. If you want to like and subscribe, you can find us on Apple, Spotify, or wherever you get your podcasts. We'd love to hear your feedback on this episode of the podcast itself. Please email us@podcastren.org. Endocrine Feedback Loop is a free service of the Endocrine Society. To learn more or to become a member member, visit the society's website at www.endocrine.org.