Lowering Hashimoto antibodies: what measurably changes

The lab report is on the table. One line is printed in bold, with a small arrow pointing up beside it. TPO antibodies: 1,240. Reference range up to 34.

You know the word behind it by now. Hashimoto's. And then what almost always happens happens: you start searching for this number and for how to make it smaller.

I understand that well. A number is something you can hold on to and follow. A diffuse feeling made of fatigue, sensitivity to cold and brain fog is not.

Still, an uncomfortable question before we talk about selenium, gluten and vitamin D. Suppose this value reads 620 a year from now. What exactly would you have gained? The honest answer to that is the core of this article.

The question before the question: what does an antibody titer actually say?

Many people with Hashimoto's know this feeling: the titer turns into a competitive sport. Written down, compared, posted in forums. It becomes the grade the body hands out. Except that it measures something other than most people believe.

TPO antibodies are directed against thyroid peroxidase, the enzyme the gland uses to build iodine into its storage protein. Thyroglobulin antibodies are directed against that storage protein itself. Both show that the immune system is occupied with thyroid tissue. Why it does that I have told in Hashimoto: why the immune system attacks. Here I take it as given.

The question in this text is a different one. It runs: what measurably changes this number, and what changes along with it?

Positive versus negative is a meaningful distinction

This is exactly where everyday logic slips. A yes or no piece of information turns into a scale in your head. Positive becomes a lot, a lot becomes bad. For this scale there is no recognized grading. No laboratory tells you from which point a value counts as severely elevated. That is not an oversight, the data simply are not there.

What the height has to do with how you feel

A second hint comes from a field where you would not expect it, fertility medicine. In the British TABLET trial 19,585 women were tested for TPO antibodies. 952 antibody positive women with normal thyroid values were randomized to levothyroxine or placebo, from before conception to the end of pregnancy. The live birth rate was 37.4 versus 37.9 percent, no difference, and none for miscarriage or preterm birth either [PMID 30907987]. A risk that is defined via antibodies cannot therefore be defused automatically by turning a different screw. The marker marks something. That does not yet make it the lever.

If you feel awful and your antibodies are high, both things are real. Only the second finding is not automatically the explanation for the first. Anyone who feels bad despite normal values will find more in Normal values, symptoms anyway.

And now you know why I begin with this preliminary question instead of immediately writing down a list of antibody lowering measures.

Why antibodies fall on their own, and what that does to your before and after

Imagine two people. Both have Hashimoto's, both start at a TPO value around 3,000. One of them changes nothing for a year. The other changes their diet, takes selenium and vitamin D and avoids gluten. After twelve months both measure again, and in both the value is considerably lower.

Which of the two has seen the effect of their measures? That is the methodological problem of this whole field, and it is well documented.

For every personal account online that means: there is a powerful natural comparison arm here. Without a control group, a fall in titer is not a statement about effect but an observation within the expected range.

And then there is the laboratory

A second confounder is mentioned even more rarely. Antibody assays are not calibrated uniformly, different manufacturers measure the same sample differently. If your practice changes laboratory or the laboratory switches its test system, a value can jump by several hundred units without anything having changed inside you. A comparison across the years is therefore only halfway meaningful when it was produced in the same laboratory with the same method. And even then a biological range of variation remains.

And now you know why I read before and after stories on this particular marker so cautiously. Not because people are not honest. But because the marker moves anyway.

Selenium: the most reproducible effect, and the trial that shifted everything

If you ask for advice in a Hashimoto group, selenium arrives within two minutes. It does have a decent body of trial data. Only that data shows something other than its reputation suggests.

Step one: why selenium is a candidate at all

The thyroid works with an aggressive tool. To build iodine into its storage protein it needs hydrogen peroxide, the same molecule used to bleach hair. Useful and risky at the same time. So that the gland cell does not damage itself, it keeps protective enzymes ready, and some of them contain selenocysteine, an amino acid with selenium at its center. The same group of substances sits in the deiodinases, the enzymes that convert T4 into the more active T3, see From T4 to T3.

Step two: what the meta-analyses find

The heterogeneity of 90 percent is almost as important as the effect. The individual trials found very different results, and the average is then an arithmetic quantity, not an expectation for the individual case.

Look more closely and it gets more restless. An analysis of 16 controlled trials found TPO antibodies lower on levothyroxine at every time point, without levothyroxine only after three months [PMID 27702392]. A Chinese analysis from 2023 arrived at the opposite time pattern [PMID 37335715]. Another one from 2025 covering 21 trials found a difference by form: selenomethionine showed more consistent titer effects than sodium selenite or selenium yeast [PMID 40898469].

When two meta-analyses of the same field cannot agree on when an effect appears, that effect is not robust. The closing sentence of the Danish group from 2016 is the most candid in the whole field: whether these changes are connected to clinically relevant quantities remains to be shown.

And there is a formal grade for it. A group around Yishen Wang assessed the existing systematic reviews in 2023, with AMSTAR-2 and GRADE. Of six reviews only one counted as methodologically high quality, and for the best known finding the certainty of evidence reads very low [PMID 37513612]. The sentence that selenium can lower TPO antibodies therefore describes a real finding which at the same time is formally documented with very low certainty.

Step three: the gap that stood open for twelve years

In 2013 Cochrane looked at the field. Four randomized trials with 463 participants, a meta-analysis was not feasible because of the heterogeneity. The most interesting finding was not in the numbers but in a gap: in none of the four trials was health related quality of life assessed at all. Selenomethionine lowered TPO antibodies, sodium selenite not significantly [PMID 23744563].

For twelve years nobody knew whether people with lowered antibodies are doing any better. Then came CATALYST.

Fairness requires the counterpart. A meta-analysis from 2010 found across three small trials a higher chance of reporting an improvement in mood or wellbeing on selenium, risk ratio 2.79 [PMID 20883174]. That number still stands behind many recommendations today. It rests on non standardized self reports in small groups. CATALYST later asked the same question with a validated instrument and 412 people.

The safety question that is usually missing from guides

With selenium there is a narrow optimum. The review by the group that also ran CATALYST puts it in such a way that unwanted effects can occur at both ends of intake [PMID 32001830]. More is not automatically better here.

At the same time the supply situation in Europe is a genuine issue. A systematic review of 19 intake studies and 48 status studies describes a widely suboptimal selenium status, essentially because of the selenium content of the soils [PMID 25734564]. From my point of view that is the one legitimate reason to think about selenium in Germany. Not the titer, but an actual undersupply. In more detail in the spoke Selenium, zinc, iron and vitamin D.

An Italian survey of 778 specialists arrived at even higher numbers: 79.4 percent prescribed selenium in chronic autoimmune thyroiditis [PMID 27843806]. I do not want to smooth this tension over. Doctors who know the trial data also reach for selenium. That is not down to missing information, but to the fact that a substance with plausible biology, a low price and good tolerability feels like a defensible attempt. That logic has its justification. It is just not an evidence argument.

And now you know why with selenium I ask about supply status first and not about the antibody value.

Gluten free without celiac disease: what the small studies show and what they do not

Hardly any piece of advice is passed on as self evidently in Hashimoto's as this one: drop the gluten. In some groups it counts as a basic requirement. And as so often there is a good kernel in it and a lot of fog around it.

The solid part: celiac disease is more common in Hashimoto's

The shaky part: what the intervention studies show

Does a gluten free diet change the antibodies in people without celiac disease? The most cited paper on this comes from Robert Krysiak in Katowice. 34 women with autoimmune thyroiditis, 16 of them ate gluten free for six months, 18 did not. In the diet group TPO and Tg antibodies fell, in the control group they did not [PMID 30060266].

This study turns up in almost every German guide. One limitation is almost never cited with it: all participants had positive anti transglutaminase antibodies, that was the inclusion criterion. So it is not a study on Hashimoto's without celiac disease, but on Hashimoto's with a serological suspicion of celiac disease.

And now it gets interesting, because the two existing meta-analyses contradict each other.

The cleanest formulation of the topic comes from a recent mechanism review: in celiac disease the role of gluten is established, in Hashimoto's gluten is better understood as a possible modifier, not as a proven universal trigger. What is discussed are gut dysbiosis, a more permeable epithelial barrier, cross reactivity and epigenetic regulation. For Hashimoto's without confirmed celiac disease the evidence is not sufficient for a general recommendation [PMID 42147111]. The barrier question beyond the thyroid I have described in Gluten and gliadin without celiac disease and in Leaky gut and intestinal permeability.

A change of perspective to close

What nutrition can do in Hashimoto's overall, beyond the antibody question, is in Nutrition in Hashimoto's, and elimination diets I discuss in Paleo, AIP and autoimmune conditions.

And now you know why with gluten I ask about the celiac investigation first and only then talk about styles of eating.

Vitamin D: good prevention data, thin data on the course

Vitamin D has had a steep career. It is involved in almost every immune question, it is cheap, and at our latitudes the levels are low in many people in winter. Even so a distinction is worth making that almost always blurs online: the one between prevention and course. Two questions, two answers.

The prevention question

This number is quoted often. The follow-up almost never.

The question of the course

And if Hashimoto's is already there? Here the pattern from the selenium section repeats itself. A meta-analysis from 2023 across twelve studies and 862 people found clear effects on the antibodies, TPO-Ab with a standardized mean difference of minus 1.084, while for TSH the confidence interval included zero. One detail stands out that guide texts practically never mention: the strongest antibody effect came from calcitriol, the prescription only active metabolite, not from the vitamin D3 off the drugstore shelf [PMID 38206745].

A second meta-analysis across six studies and 258 people likewise found a decline in TPO antibodies, but no difference for Tg-Ab, TSH, fT3 or fT4. The conclusion of the author group is remarkably direct: in people with Hashimoto's, vitamin D is not associated with thyroid function [PMID 34981556].

My conclusion for practice: bringing a low vitamin D level up into a sensible range makes sense for many reasons, from bone to muscle strength to immune regulation. As a tool for lowering antibodies it is, in my view, as unsuited as selenium, and for the same reason.

And now you know why the VITAL number so often sits in the wrong place in guides. It belongs to the emergence, not to the course.

Myo-inositol plus selenium, and what "not established" concretely means

One combination has a good reputation in forums and in some practices: myo-inositol together with selenium. The idea behind it is elegant, so I treat it properly instead of dismissing it.

Myo-inositol is a sugar alcohol that sits in cell membranes and takes part in signal transmission. Interesting for the thyroid because it is described as a second messenger of TSH. Simplified: when the signal from the pituitary arrives at the gland cell, it needs a messenger on the inside. Mechanistically understandable. And then comes the data.

The second paper from the same group covered 86 people and had no control group at all [PMID 28293260]. Given the documented natural decline over time, that finding is hard to interpret without a comparison arm. The third and oldest paper from 2013 was double blind and randomized, but very small [PMID 24224112]. Three papers, one research group, one country, proximity to a manufacturer, no independent replication since 2013. That is not a verdict on the substance, it is a description of the evidence level.

And now the uncomfortable part: what is not established

In many guides substances appear in recommendation lists for which there is no human study with an antibody endpoint in Hashimoto's. This is not about the people who offer them. It is about the data, and at these points the data are empty.

One exception leads into neighboring territory: the question of environmental exposures and autoimmunity. That is a topic of its own with data of its own, and I have handled it separately in Heavy metals and Hashimoto's. It does not belong in a list of antibody lowerers, but in the question about the history.

And now you know why with long recommendation lists I always look for the source citation first.

Low dose naltrexone: why it runs in the scene and not in the trial data

In autoimmune forums you sooner or later meet an abbreviation: LDN, naltrexone at a very low dose. Naltrexone is an opioid receptor antagonist from addiction medicine. The idea behind the low dose: a brief, incomplete blockade of the receptors could trigger a counter regulatory response and intervene in immune regulation along the way, among other things via microglia and toll like receptors.

That sounds plausible, and people report that they feel better with it. Even so the trial data look different from what the reputation suggests.

On top of that comes a formal point: use at a low dose is an off label application. In certain situations that is legitimate, but it requires explanation, medical supervision and a clear agreement about how you will notice whether it brought anything.

I write this without scorn. The author group itself considers a pragmatic role in treatment resistant courses conceivable, provided the experimental character is clearly explained. What I do not want is a back door. The trial data have looked, they just did not find much.

And now you know why LDN gets its own section in this article but no place in a list of recommendations.

What changes more in practice than any titer

If you have read this far, you know the pattern. It turns up in every section, with very different substances. So here it all is side by side.

The left hand column of numbers is the one people wrestle over online. The right hand one is the one that describes your life. In no row do the two move together.

What the guideline makes of it, and why that is understandable

The German S2k guideline on an elevated TSH value in general practice draws a clear consequence from this. In latent hypothyroidism a single determination of TPO antibodies can be made in order to clarify a suspicion of Hashimoto's thyroiditis. And then comes the sentence many people stumble over: a repeat measurement should not be carried out, because no consequences for further treatment follow from a TPO trajectory. The same guideline also states that there is no evidence for a patient relevant benefit of trace elements, herbal preparations or food supplements in hypothyroidism and Hashimoto's [DEGAM S2k, AWMF 053-046].

I understand that this feels like a refusal. The logic behind it is good nonetheless. A measurement from which no action follows produces three things: worry, follow-up appointments and costs. Guidelines therefore consistently examine patient relevant endpoints, not markers. That is not a lack of interest in you, it is an attempt to protect you from pointless worry.

What a functional view asks in addition is something else. It reads the antibody load as an expression of immune activity and asks what accompanies that activity. Not in order to turn it into a target, but in order to understand what state the system is in right now. Both questions are legitimate, they just have different aims.

The European professional society places the antibodies in a similar way: initial classification and risk assessment where TSH is elevated, not a treatment target [PMID 24783053]. And the American guideline names as a research need that better biomarkers of euthyroidism than TSH alone would be required [PMID 25266247]. So the guideline world does not find the current state satisfying either. That is exactly where a functional view can connect, without contradicting.

What I look at instead

I see nutrients as cofactors, not as active agents. They make metabolic steps possible, they do not force them. More on that in From calorie to energy.

Pregnancy and wanting to conceive

Here separate, pregnancy related reference ranges apply. The most common cause of subclinical hypothyroidism in pregnancy is autoimmune thyroiditis, and both it and isolated hypothyroxinemia are associated with less favorable obstetric courses [PMID 25114871]. That is a framework of care for medical hands, not for self experiments.

For the antibody question TABLET remains the most important data point: in euthyroid, TPO positive women levothyroxine did not change the live birth rate. On selenium supplementation in pregnancy the German guideline records that no recommendation can be derived because the evidence is weak. On the cycle see Thyroid and female hormones.

When something belongs in medical hands

A finding at the margin, explicitly not an option

Why this is not a recommendation despite impressive numbers: an operation cannot be blinded, and the main endpoint was a subjective questionnaire. On top of that come a lifelong dependence on hormone replacement and the risks of any procedure on the neck. I cite the study exclusively as a hint about mechanism. Whether an operation on the thyroid comes into question in an individual case is decided in a specialist assessment and never via an antibody value.

One last thought, which I mark clearly because it is not a study finding. In the anthroposophic way of looking at things the thyroid is connected with the realm of speaking out, that is with the question of what a person brings out and what they hold back. That is a level of interpretation from clinical tradition, not physiology and no substitute for it. It stands next to the laboratory values, not in their place. In more detail in the spoke The thyroid seen anthroposophically and in Anthroposophic medicine.

And now you know why in my consultations I rarely ask how high your TPO value is, and almost always ask how you slept.

Common questions about Hashimoto antibodies

Autoimmunity does not stop at the thyroid

An immune system that occupies itself with the body's own tissue rarely does so for one single reason. From here several paths lead onward, and none of them ends in an antibody number.