Can a normal TSH rule out thyroid disease?

If you are well-informed by current thyroid science, you already know the answer to the question in the title. “No. A normal TSH cannot rule out thyroid disease.”

Since the 1990s, physicians have been taught to use simplistic category-based interpretations of TSH (and Free T4) as “in or out of range” to classify people as “euthyroid” or with degrees of “subclinical” or “overt” hypo- or hyperthyroidism.

However, a recent scientific article asked a similar question in its title: “Does TSH Reliably Detect Hypothyroid Patients?” the answer was also “no.”

Normal TSH values may not rule out patients that are hypothyroid.

It is possible that a large number of hypothyroid individuals are missed using TSH as a screening tool, which is a problem for the patient, who remains hypothyroid, and to the healthcare system as a whole.”

(Ling et al, 2018)

Ling’s article wasn’t written by a few physicians who take pleasure in being contrarians. It was published by 13 coauthors in a peer-reviewed journal called Annals of Thyroid Research in 2018.

One of its many authors is Jacqueline Jonklaas, with 106 publications listed in the Scopus database, and lead author of the 2014 American Thyroid Association’s Guidelines for the Treatment of Hypothyroidism. Another coauthor is Steven J. Soldin, who has his name on 299 publications, many of them on the topic of TSH and thyroid hormone assays. Listen to their team’s wisdom:

“Although measurement of TSH is a convenient screen for thyroid function, it is influenced by many factors which may affect its overall reliability.

We believe thyroid function should be assessed by more than a single test.”

(Ling et al, 2018)

In other words, TSH is a “convenient screen” but it can be “unreliable” as an indirect measurement of “thyroid function.” To demonstrate, Ling’s article discusses two cases of hypothyroidism with a normal TSH.

Ling and colleagues aren’t the only scientists who have been pointing out that a normal TSH cannot rule out thyroid dysfunction.

Medical overconfidence in the TSH test as a “gold standard” for diagnosis is holding back the field of thyroid science, according to Rudolf Hoermann, a thyroid scientist with 175 publications to his name:

Over-reliance on TSH as a gold standard has long impeded the advancement of the field, since the first doubts were raised and disagreements emerged on the setting of the reference intervals.”

(Hoermann et al, 2016)

The main problem is not the sensitivity of our TSH assay technology, which is quite precise, although some problems remain with specificity and inter-assay harmonization. Instead, the main problem is with a paradigm that worships TSH.

A lot of TSH-based screening and diagnosis today is based on a misunderstanding of what it means for the TSH to be a “sensitive” test. It’s also based on an outdated, simplistic model of the hypothalamus-pituitary-thyroid (HPT) axis and the “log-linear” TSH-FT4 relationship.

These beliefs and misunderstandings have been ingrained over many decades. They now support restrictive thyroid screening policies that claim to save health care dollars. They have led to overconfidence that the TSH concentration and its reference interval is the best and only framework for interpreting thyroid gland health and thyroid hormone status. It’s easy to find overblown claims that a normal TSH can “rule out” thyroid dysfunction (such as Davis & Philippi, 2022).

Some of the best thyroid scientists have now informed us that each person’s pituitary TSH secretion rate in response to FT4 and FT3 is enhanced or inhibited by many nonthyroidal factors besides adequate iodine supply. Not only common drugs like metformin or steroids, but also a severe chronic illness, one’s age, thyroid antibody status, or poor TSH quality, or an overactive or underactive pituitary, can cloak primary thyroid dysfunction within the normal TSH reference range (Ling et al, 2018). Likewise, unconscious exposure to endocrine disruptors that distort TSH secretion may be either regional or individualized.

In this article, I gather together many lines of scientific evidence that illustrate that a normal TSH can be present in an untreated individual with a clinically-significant thyroid dysfunction, even when there is no pituitary or hypothalamic dysfunction.

To patients who have experienced misdiagnosis of “euthyroid” status due to their normal TSH, it’s a delayed and bittersweet victory to see how all these leading researchers — and many more — are moving into a new era of evidence-based TSH suspicion about the absence of thyroid dysfunction within the TSH reference interval.

Now, patients are waiting for physicians, endocrinologists, and health care systems to pay attention to the enlightened thyroid scientists. Who is willing to remove the institutionalized cognitive barriers to seeing thyroid dysfunction within the TSH reference interval?

Table of contents

Series summary: Cognitive barriers

This post is part of a series on the cognitive barriers to the analysis of an untreated person’s normal-range TSH, FT3, and FT4 lab data.

Expand to view summary and links

Series links:

  1. Cognitive barriers to analyzing “normal” thyroid lab results
  2. Are normal thyroid reference ranges risk-free zones?
  3. What’s wrong with calling TSH, FT3 and FT4 “thyroid function tests”?
  4. This post — “Can a normal TSH rule out thyroid disease?”

A cognitive bias is like a preference to go through one door rather than another.

A cognitive barrier means certain doors to deeper knowledge are locked or entirely hidden from view.

Any cognitive “bias” can become a cognitive “barrier” when it prevents one from examining the medical evidence that is right before their eyes – normal-range TSH, FT4, and FT3 lab data.

In the series introduction, I provided a comparison across medical fields. I outlined how scientists and policymakers in other diseases treat “normal range” values with great skepticism. In contrast, thyroid medicine tends to cling to the clinical and diagnostic significance of its normal reference ranges.

The first cognitive barrier is the belief that “normal” range is a zone where thyroid hormones and TSH pose no health risk for any non-thyroidal disorders. For a long time, scientists have ignored non-thyroidal disease risk within the FT3, FT4 and TSH ranges because it is quite clear that very abnormal levels are riskier for untreated people than are normal levels, relatively speaking.

But science is now revealing that imbalanced yet normal thyroid hormones and TSH can negatively affect cardiovascular health, kidney health, liver health, and so on.

The second cognitive barrier is the common phrase “thyroid function test”. It narrows the mind by narrowing the medical purpose of the test and failing to mention what else it can reveal. The language focuses on one corner of the HPT axis, the corner that is most accessible to examination and easiest to blame. But analyzing two or three hormones at once enables you to gain perspective, just as using two eyes instead of one enables you to judge depth and distance.

In the hands of an intelligent analyst, the TSH, FT4 and FT3 tests uncover the fine distinction between thyroid gland health and tissue thyroid hormone status, a distinction that is too often collapsed and misunderstood as being synonymous. They also help one evaluate pituitary (and hypothalamus) function as well as peripheral thyroid hormone metabolism.

The third cognitive barrier (discussed here) is the widespread yet mistaken belief that a normal TSH has the power to single-handedly rule out almost all thyroid function disorders except in the case of so-called “rare” pituitary disorders. But once you’ve gotten rid of the first and second barriers, this barrier to knowledge is much easier to get out of the way.

Physiology teaches us that a normal TSH is only as effective as the FT3 and FT4 reveal it to be. A normal or abnormal TSH may either fail or succeed at maintaining the untreated individual’s healthy thyroid hormone setpoint.

After the discussion of all these cognitive barriers, we must confront the paradox: Why is it part of evidence-based medicine to discourage physicians from analyzing the rich evidence provided by normal-range thyroid lab data? It’s tied up in history, economy, and medical education. Some people pride themselves on being “evidence-based” when they are simply promoting or conforming to a programme of institutionalized thyroid ignorance within the normal ranges.

It’s time to stop making the upper and lower “limits” of the TSH, FT3 and FT4 ranges into the limits of medical understanding.

Not covering thyroid therapy

In this series, I focus on the screening and diagnosis of untreated individuals who still rely on a TSH-stimulated thyroid for 100% of their thyroid hormone supply.

This is not a series about the decision to treat, nor the optimization of treated thyroid patients’ hormone levels. That’s a complex topic. Many types of thyroid disorders exist and can overlap. Many treatment approaches exist. Disease and dosing will combine with genetics and nonthyroidal health conditions to alter a person’s optimal thyroid hormone setpoints.

However, TSH barriers to diagnosis are still relevant to thyroid therapy. Cognitive barriers to analyzing untreated people’s normal-range thyroid data are at the foundation of many thyroid therapy mistakes. If a doctor believes “normal” is safe and healthy in every untreated person, their cognitive barriers will also become barriers to optimizing thyroid therapy rather than merely normalizing their biochemistry.

COPYRIGHT NOTICE: Reproduction of a copyrighted article’s data, graphics, and quotations falls within US copyright law as “fair use” and within Canadian copyright law as “fair dealing.” — “A fair use is any copying of copyrighted material done for a limited and “transformative” purpose, such as to comment upon, criticize, or parody a copyrighted work. Such uses can be done without permission from the copyright owner” (Stanford University).

Illustration 1. TSH is a very poor predictor of FT4

A study of TSH-FT4 relationships conducted to Hadlow and colleagues in 2013 involved 120,403 well-screened, untreated people.

In the image below, the dotted lines surrounding the male and female medians represent the means of the upper (Q4) and lower (Q1) quartiles of TSH per FT4 level. Their reference ranges were TSH (0.4-4.0 mU/L) and FT4 (10-20 pmol/L).

In the graphic, the distances between the arrows express the wide range of human diversity in FT4 levels found at every TSH level, and the wide range of human diversity in TSH levels found at every FT4 level:

  • At FT4 10 pmol/L, TSH is likely to fall between 3.0 (Q1) and 10.0 (Q4)
  • At TSH 4.0 mU/L, FT4 is likely to fall between 9.0 (Q1) and 17.5 (Q4)
  • At FT4 20 pmol/L, TSH is likely to fall between 0.01 (Q1) and 3.0 (Q4)
  • At TSH of 0.4 mU/L, FT4 is likely to fall between 17.5 (Q1) and 27.0+ (Q4)

According to official diagnostic guidelines (Garber et al, 2012, p. 999), low FT4 levels trump TSH levels. At any TSH level in an untreated person, a FT4 below reference range immediately changes the diagnosis to biochemical “overt hypothyroidism.” If combined TSH-FT4 testing were not permitted, nobody with central hypothyroidism would ever be properly diagnosed (Beck-Peccoz et al, 2017).

As you can see, a “normal” TSH cannot exclude a FT4 value that signals thyroid dysfunction.

Isn’t it amazing that the very same high-normal FT4 level of 17.5 can be associated with a TSH of 0.4 in one person and 4.0 in another person? At a TSH close to 0.4 mU/L, only the concurrent FT3 level will determine whether a FT4 of 17.5 is contributing to a patient’s tissue hyperthyroid signs and symptoms. A T3-secreting thyroid nodule can escalate the FT3 level while TSH remains normal (Chami et al, 2014).

Anyone who uses TSH to screen for thyroid function ought to be aware of the limited predictive value of a “normal” test result.

Why does this happen in population data? This aspect of TSH-uncertainty is created by the role of TSH in biology, as I discuss later.

Illustration 2. A hypothyroid man with a normal TSH

Population statistics can describe the probability of misdiagnosis when TSH is normal, but case studies prove that a misdiagnosis is possible for individuals, and demonstrate how to avoid it.

Ling et al, 2018 provide a case study of a patient whose normal TSH was clinically suspicious and was correctly diagnosed as hypothyroid:

“Case report: This is a 76-year-old male who presented with the following symptoms: fatigue by early afternoon, constipation, dry skin, and hair loss. His laboratory values by IA [immunoassay] were normal for all thyroid tests. His TSH was 2.65 μIU/mL and his cholesterol was 250 mg/dL.”

(Ling et al, 2018)

In the 2012 American Thyroid Association guidelines by Garber and colleagues, elevated total cholesterol is an indication for TSH testing because it is well known to be “correlated with” hypothyroidism, but by itself, it is “not sufficiently specific to diagnose hypothyroidism.” Nevertheless, normalization of high cholesterol is one of many “confirmatory findings that patients have been restored to a euthyroid state” by treatment, so it was worthy of mention in light of treatment results (quoted below).

This patient’s immunoassays had revealed unspecified normal FT3 and FT4. However, the gold-standard method, mass spectrometry (LC-MSMS) assays revealed free thyroid hormones just below their reference ranges: FT3 2.1 (2.2–6.2) and FT4 1.2 (1.3–2.4).

Given the patient’s thyroid hormones and clinical presentation, he was treated on low-dose T3/T4 combo therapy.

“He was started on 5 μg/d T3 and 25 μg/d T4.

After 2 months on low-dose thyroid hormones, his FT4 increased to 1.8 ng/dL [1.3–2.4] and FT3 increased to 3.6 pg/dL [2.2–6.2] by LC-MSMS. The patient’s TSH and cholesterol dropped to 1.95 μIU/mL and 232 mg/dL, respectively.

Significantly, he described having increased late afternoon energy and improved cognitive function, and he no longer suffered from constipation and dry skin. The patient also reported that his urinary flow-rate has increased by what he estimates to be roughly 3-fold since taking thyroid hormones.

The patient’s hypothyroidism presented in the context of a normal TSH and was successfully treated by a low-dose thyroid hormone regimen.”

(Ling et al, 2018)

An improvement in biomarkers and symptoms shows that treatment was warranted. It is well known that an adult human thyroid produces much more than 25 mcg of T4 per day, and oral absorption of levothyroxine (LT4) is approximately 85% at best.

If he had had healthy HPT axis function, the low-dose T4 regimen would have merely replaced some of the T4 from his existing TSH-stimulated thyroid function and made little difference to the numbers.

We are not told of the reason why his “hypothyroidism presented in the context of a normal TSH,” but the authors suggest excess glucocorticoids and/or endogenous ACTH-stimulated steroid hormones were likely involved:

The administration of glucocorticoids decreases TSH levels [6,13–17]. ACTH stimulation leads to an increase in many steroids in healthy patients [24], which is the likely cause of the decrease in TSH we observed in this study.

(Ling et al, 2018)

Ideally, the true cause of his TSH decrease, whether biochemical interference or “central hypothyroidism,” should be investigated. It could be added to this patient’s medical record so that future physicians, including emergency physicians, are aware that his TSH during thyroid hormone treatment may become an inappropriate a guide to his thyroid therapy or hospital care.

There are many other illustrations of ways that thyroid dysfunction manifests itself within the normal range. These will need to be covered in separate posts —

  1. Age bias may hide hypothyroidism under a normal TSH
    1. Pediatric and teenage TSH, FT4, and FT3 levels
    2. Age, sex and TSH-FT4-FT3 relationships: Advanced lessons
  2. Antibody bias may hide thyroid dysfunction under a normal TSH (Coming soon!)
  3. Drugs like metformin may hide thyroid failure under a normal TSH (Coming soon!)
  4. Severe illness can distort TSH and inhibit thyroid function (Coming soon!)
  5. Nephrotic syndrome may hide hypothyroidism under a normal TSH (Coming soon!)
  6. Functioning thyroid nodules and T3-toxicosis may hide under a normal TSH (Coming soon!)

The rest of the post talks about how the current scientific understanding of TSH technology and TSH in biology makes these apparent paradoxes possible.

TSH assay technology vs. TSH biology

Why have so many intelligent misunderstood that a normal TSH can’t rule out thyroid dysfunction? Partly because “sensitivity” can mean so many things when a person praises the “sensitivity” of the TSH test.

“TSH has gained a dominant but misguided role in interpreting thyroid function testing in assuming that its exceptional sensitivity thereby translates into superior
diagnostic performance.

(Hoermann et al, 2016)

When people express overconfidence in the sensitivity of the TSH test, these three concepts are often confused as well as oversimplified:

  1. The high quality technology of the TSH assay (laboratory test), despite a few remaining weaknesses, provided that the reference range is accurate for the assay and population.
  2. The biology of TSH hormone concentrations, which are co-regulated by many variables beyond thyroid hormones.
  3. The clinical predictive value of the isolated TSH test as a screening test in the process of diagnosis of an untreated individual, i.e. the likelihood of a positive (abnormal) TSH result to predict a true positive diagnosis of hypo- or hyperthyroidism once it is confirmed by more definitive testing. This also involves a certain degree of likelihood that a normal (or abnormal) TSH-only screening will yield a false negative (or false positive) result.

Aspect #1 is a foundation for interpreting #2, and problems with both 1 and 2 can undermine #3, diagnosis. In this post, I won’t be dealing with aspect #3 in detail, but once one understands the two foundations of TSH technology and biology, one will understand better how a normal TSH can yield a false negative prediction of euthyroidism.

1) TSH assay technology

Many of the strengths of the TSH test were revealed in comparison to its historical competitors, such as Total T4 testing and the TRH-stimulated TSH test.

In thyroid history, Robert Utiger can reasonably be called “the father of the TSH test” given his many decades of research and advocacy for this test. He was the one who repeatedly wrote about TSH being “exquisitely sensitive” to thyroid gland status and circulating thyroid hormones.

In the late 1980s, the “sensitivity” of the TSH assay had to do with the greatly improved ability of TSH to differentiate normal subjects from hyperthyroid subjects. They previously had to stimulate pituitary TSH secretion by injecting TRH hormone, and the amplified TSH level over time would reveal the individual’s thyroid hormone status according to the pituitary.

“Until recently, it was not possible to distinguish subnormal from normal serum TSH concentrations reliably. The sensitivity of the available TSH assays was about 0.5 to 2 μU/ml, such that serum TSH concentrations were undetectable in some normal subjects. …

Now, as a result of recent developments in immunoassay technology, the sensitivity of TSH assays has been improved considerably …. The sensitivity of these assays is such that serum TSH concentrations as low as 0.01 to 0.02 μU/ml can be detected.

As a result, TSH is detectable in the serum of all normal subjects, and patients with subnormal as well as those with undetectable serum TSH concentrations can be identified.”

(Utiger, 1988)

As the TSH test was refined in 1987-1988 to its third generation of technical sensitivity, it won the competition with other tests used during a former era of medicine. Many joined Utiger in a chorus of praise for the clinical utility of this improved test (Toft, 1988; Saller et al, 1998).

Now listen to Utiger’s guarded thoughts on TSH-first testing policy, back in 1988:

Should serum TSH measurement become the initial test used to evaluate thyroid function, rather than measurements of serum total and free thyroxine [T4]?

This proposal has several attractive features, as long as it is recognized that the key word is “initial.”

A normal serum TSH concentration is stronger evidence against the presence of thyroid dysfunction than is a normal serum thyroxine [Total T4] concentration, in both ambulatory and hospitalized patients, because TSH measurements identify those patients with subclinical hyperthyroidism or hypothyroidism, whereas serum thyroxine measurements do not.

(Utiger, 1988)

And as you can see, Utiger emphasizes that TSH may be useful as the “initial” test and is not the sole, isolated test. Our generation has thrown that limitation out the window in the interest of cost-savings.

Providing inconclusive yet relatively “stronger evidence” of the likelihood or degree of thyroid dysfunction is something that the isolated TSH level can do. That is certainly not something the isolated Total T4 level can do.

The “subclinical” TSH-based biochemical categories have turned out to be more confusing than Utiger had ever realized. Not even the best endocrinologist can say conclusively what’s going on in the subclinical zones with only a TSH result.

Between the TSH normal range and its immediate outlying regions, an analyst requires the TSH-FT4 relationship, and additional FT3 data and clinical context, to distinguish the biochemical features of healthy thyroid function from those of compensated or uncompensated thyroid or pituitary dysfunction in the untreated individual. (I’ll address successful “compensation” later below.)

But we are no longer in the 1990s. Today, health care systems are no longer struggling with the fallibility of Total T4 and indirect estimates of Free T4 by using outdated, imprecise methods. In addition to a further refined TSH technology, we now have cheap, high-quality FT3 and FT4 tests as well as slightly more pricey thyroid antibody tests. The latter are far more useful than Total T4 to confirm the results of a TSH screening test.

Today, in 2022, when a biochemist says that a TSH assay is highly sensitive and specific as a technology, they are not comparing which test in isolation, the TSH, FT3 or FT4, is “stronger evidence” of thyroid dysfunction. They simply mean that the test tells the biochemical truth — it reports the analyte, the whole analyte, and nothing but the analyte, even at low concentrations:

  1. Capable of detecting all the TSH in a blood sample,
  2. Capable of accurately measuring very low and very high levels of TSH, and
  3. Capable of ignoring TSH-like molecules that are not bioactive TSH.

Issues #1 and #2 are pretty much resolved today. That’s old news.

Despite the TSH test’s improvements over the years, it still fails at criterion #3, ignoring TSH-like molecules that are not bioactive TSH.

“Measurement of Thyroid Stimulating Hormone (TSH) concentrations by immunoassay is used by the majority of clinical laboratories to assess thyroid function. These assays are subject to a number of [assay] interferences, including

• biologically inactive isoforms of TSH,
• heterophilic antibodies and
• biotin supplements.

(Ling et al, 2019)

The “biologically inactive isoforms of TSH” mean that some TSH molecules secreted by the pituitary, such as “macro-TSH” molecules, may fail to stimulate TSH receptors in the thyroid gland. The larger molecule size of inactive TSH isoforms also can also inflate TSH levels because they clear more slowly from blood (Hattori et al, 2018; Katoya et al, 2017). They can inflate a subnormal TSH into the normal range in a hyperthyroid person, or they can make a normal concentration of TSH impotent at the task of thyroid stimulation.

Heterophilic antibodies have been a TSH assay interference since the 1980s. In addition to biotin interference, scientists have discovered anti-streptavidin antibody interference and anti-Ru interference in TSH results (Favresse et al, 2018).

A fourth issue not mentioned above plagues TSH tests, FT3 tests and FT4 tests — inter-assay variability and reference interval variability.

In 2018, another group of scientists discovered that two popular TSH assays disagreed about the concentrations of values in the upper normal range.

The discrepancy went beyond their placement of the upper limit:

For TSH values 2.0 to 6.0 mIU/L, a difference between assays of -2.33 in one direction and 2.8 in the other direction is significant enough to misclassify one patient as “euthyroid” and another as “subclinical hypothyroid.”

“Our results demonstrated the TSH measurement has a poor correlation in a range where more precision is necessary—the euthyroid and the subclinical hypothyroid range.

“the disagreement observed in some cases is
randomly distributed in the normal upper range and close above this value, which we consider a clinically delicate and unstable range of TSH values.”

(Da Silva et al, 2018)

When two TSH assays disagree about whether your TSH result is 3.0 or 5.5, which one is correct? Even if the two assays’ reference limits were the same, the fact remains that they disagree on the TSH concentration in the same blood sample. The discrepancy widens as TSH rises within range.

A mere -1 to +1 SD difference that would otherwise be a minor assay discrepancy can become major barrier to diagnosis because many jurisdictions have made a normal TSH shut the door to thyroid hormone testing.

Is the symptomatic hypothyroid patient lucky to live in a province with a Roche assay and a reference limit of 4.0? They may gain access to a FT4 test and a fuller diagnosis of their thyroid function. Too bad for the hypothyroid person living in an Abbott assay region, or any region that has an inflated upper reference interval.

Unfortunately, there is no more accurate “gold standard” TSH assay method to which all TSH assays should be calibrated. The best that biochemists and endocrinologists can do is to promote “harmonization” among the existing TSH assays, as recommended by an international committee (Thienpoint et al, 2017).

But TSH assay harmonization can be done very badly without following the guidelines and cautions set by the international committee, and it is not a cure-all:

“However, even after harmonization minimizes inter-method differences, it remains to be determined to what extent universal ranges would be impacted by other factors such as age, ethnicity, and iodine intake.”

(Spencer, 2000/2017)

Those “other factors” cannot be controlled by the TSH assay technology because they are part of the uncertainty caused by TSH biology.

2) TSH in biology

Due to the complex role of TSH in biology, confidence in the TSH assay’s technical or analytical sensitivity does not logically translate to confidence in a normal TSH’s independent negative predictive power to rule out thyroid disease.

As shown in the first illustration of the TSH-FT4 relationship above (Hadlow et al, 2013), a normal TSH is not a perfect predictor of a normal FT4, much less a normal FT3.

Biological factors that distort TSH levels are included in what clinical biochemists call “pre-analytical” factors — variables that occur before the assay equipment analyzes a blood sample. You can’t blame the laboratory or TSH assay technology if your unique biology has distorted your TSH.

The list of “other factors” that interfere with TSH levels has been getting longer and longer. The best scientific articles will list more than five factors but still can’t encompass them all:

“The concentration of serum TSH is affected by a number of non-thyroid hormone factors, such as

• pregnancy,
• age,
• sex,
• exercise,
• individual thyroid hormone set points,
• the timing of levothyroxine administration,
• steroids, and
• drugs.

Additionally, TSH may also undergo seasonal and diurnal fluctuations.”

(Ling et al, 2018)

Oversimplified models of the HPT axis

Simplistic beliefs from the 1980s and 1990s regarding pituitary TSH (in response to T3, T4 and TRH) had to be questioned as it was discovered that TSH secretion was modulated not only by glucocorticoids, illnesses, and aging (these factors were known before 1990), but also by many other metabolic influences such as genetic polymorphisms, sleep schedule, cold exposure, fasting, obesity, and exercise (Joseph-Bravo et al, 2015).

But despite the advancement of thyroid science since the 1990s, many physicians are still taught an outdated, simplistic closed-box model in which TSH responds only thyroid hormones, primarily Free T4 and secondarily Free T3 (only the “free” fractions can be carried by transmembrane transporters into cells).

In the stereotypical HPT axis model, all that we see are the hypothalamus, pituitary, and thyroid talking to each other in a closed-loop system, with the huge hypothalamus “on top” in dominant position, and the pituitary’s magnified lobes dangling over an uncomplicated smaller blob called a thyroid.

This model attributes to the pituitary alone the ability to self-regulate internally through “autocrine and paracrine regulation.” Fortunately, Ortiga-Carvalho and colleagues’ 2016 article about the HPT axis revealed many of the complexities omitted from this basic model.

The simple visual models fail to acknowledge that TSH concentrations are not just a result of the pituitary’s response to hypothalamic TRH, circulating FT3 and FT4, but also many other factors in human biology that are omitted from most HPT axis graphics.

Logic problems also exist with the closed-loop HPT axis model:

  • If it were true that the hypothalamus TRH and pituitary TSH could only respond to changes in circulating FT4 and FT3 and nothing else, then the hypothalamus-pituitary unit would be merely a robotic slave to thyroid hormones, not truly a regulator of thyroid function.
  • If it were true, their relationship would be a closed loop of reciprocal regulation. You could turn the graphic upside-down like a playing card. A jack of spades upside-down is still a jack of spades. But nobody likes the idea of a FT3 and FT4 level accusing the TSH of misbehavior.
  • If it were true that TSH regulates FT4 and vice versa, and that FT4 regulates TSH without any biological interference, you could plot a perfectly log-linear TSH-FT4 relationship on a graph. But it just isn’t true.

The TSH-FT4 relationship is logarithmic, but not linear.

An old-fashioned view of a “log-linear” relationship is often cited in articles that aim to make physicians feel confident in TSH-first testing policies (such as Mandell, 2016).

But a linear trendline is a poor fit to the population data, as shown by Brown and colleagues (2016).

Because of all the “other factors” in biology that can interfere with TSH secretion, the logarithmic relationship of TSH to Free T4 in large populations is not “linear” but rather “double negative sigmoidal.”

The wobble between TSH and FT4 shows that the complete picture is more complex than the relationship between these two hormones.

In the middle of the graph above, at the point of population homeostasis, the paradox is that TSH hardly varies while FT4 changes.

Has TSH become insensitive to FT4 variation at mid-range? No. That is the collective effect of thousands of unique, individual healthy setpoints in the mid-normal FT4 range and the variable effect of TSH on individuals’ FT3 levels.

Scientists have known since the early 2000s that the difference between any two healthy individuals’ TSH (intra-individual variation) is far greater than the difference in TSH levels over 12 months within one of those individuals (intra-individual variation) (Andersen et al, 2002, 2003).

The difference between individual TSH and population TSH has been clearly depicted by Hoermann & Midgley (2012):

Keep in mind, therefore, that there is a difference between the two-dimensional TSH-FT4 axis of any single individual and the composite relationship between two hormone medians within a population.

The population data is a background for the individual, not the judge of the individual’s thyroid status.

“The narrow TSH within-person variability and low (< 0.6) index of individuality (IOI) limits the clinical utility of using the TSH population-based reference range to detect thyroid dysfunction in an individual patient.”

(Spencer, 2000/2017)

Next, keep in mind that the TSH-FT4 relationship in Brown’s graph is like an air-brushed model on the cover of a glamour magazine, or like the composite image of a human face generated from many unique, distinctive human faces.

In 2017, Kent and Hayward wrote an excellent article describing the misleading effect of large population data sets in clinical trials. They used an art installation of human portraits to make their message clear:

“In the year 2000 artist Chris Dorley-Brown took 2,000 digital photos of people living in the small town of Haverhill in Suffolk, England, and then used software to merge these photographs, step by step, until all 2,000 had been blended into one image.”

(Kent & Hayward, 2017)

“In the modern clinical trial, the responses to treatment of thousands of individuals are typically summarized in a single number in the same way the center photograph represents all the other individuals. As the data are averaged, important individual differences are lost.”

(Kent & Hayward, 2017)

Large population models create a picture of orderliness by blurring human diversity and ignoring outliers. Can we use the image in the center to judge one person on the periphery as normal or abnormal?

The population depicted above in Brown and colleagues’ scatterplot and trendlines are screened carefully so that the graph depicts, as much as possible, uncomplicated human TSH-FT4 relationships. They are supposed to exclude people with thyroid diseases and people who dose thyroid medications. Therefore, even the “double negative sigmoidal” is not a universal, immutable “law of TSH-FT4 hormone relationships,” but what happens when there is no disturbance breaking down that relationship.

  • A properly-screened TSH “reference population” is not dosing TSH-lowering drugs, and not burdened with common yet severe chronic illnesses.
  • The TSH reference population is not a focused sub-population of pregnant people, very young children, octogenarians, or people with Down Syndrome.
  • A study of TSH-FT4 response will differ in an extreme Canadian, arctic or desert climate based on the time of year the blood was sampled.

The population-based reference ranges are simply tools in the broader diagnostic process. Physicians should not be trying to see how a patient’s TSH, FT3 and FT4 results conform to the norm, but should be trying to understand how and why the individual deviates from the norm in ways that might disturb FT3 and FT4 supply to tissues beyond the pituitary.

Some wobbles in TSH-FT4-FT3 relationships are healthy and natural. Some are caused by thyroid, pituitary, or metabolic pathology. Medicine involves the art and science of discerning the natural from the pathological. Of course it’s not easy, but people want thyroid diagnosis to be easy.

I do not believe a person is ready to interpret what is going on within the “normal” TSH range until they are truly humbled by the full picture of the complexity of the HPT axis, so I must provide a couple more visuals.

Two more inclusive models of the HPT axis

Here is a visual that includes “other factors” that influence TSH secretion that are not named in the lists above: Leptin, dopamine, estrogen, and four factors within the hypothalamus.

The image looks more complicated, like a subway map or a circuit board, but in the original source, the image above was titled “Simplified ensemble regulation of TSH secretion” (Roelfsema & Veldhuis, 2013).

It is simplified because it does not include many factors discussed by other authors.

The pituitary is now a mediator of various incoming signals and one outgoing signal to the thyroid. The hypothalamus is broken into sub-components that receive and give signals.

The word “liver” underneath the thyroid in the image is misleading. It should be replaced with “peripheral tissues.” The only time “liver” is mentioned in Roelfsema & Veldhuis’s article is in a brief passage that explains the peripheral conversion of T4 to T3 by enzymes “in different organs, eg, liver, brain, and pituitary.” The liver’s deiodinase type 1 (D1) enzyme does not contribute the majority of the body’s T4-T3 conversion per day (Maia et al, 2011).

The arrow pointing down from “liver” leads nowhere… and that’s why we need a model that includes peripheral and thyroidal deiodinases enzymes that convert T4 and T3 to other metabolites:

Deiodinase enzymes co-regulate the HPT axis

At the turning points of TSH adaptation, where TSH takes a swift upturn or downturn in Brown’s sigmoidal TSH-FT4 relationship graph, are powerful neglected factors beyond the FT4 hormone level that can’t be portrayed in a two-dimensional graph with an X and Y axis.

FT3 hormone is on the unseen “Z-axis.” All three axes are graphically represented only once, that I know of, in a rare cube-like three-dimensional model by Hoermann and team (2016), not shown here.

The major missing variable in the visual models of the HPT axis is the flexible rate of T4 conversion to T3, and the extremely flexible ratio of T3/T4 secretion from the thyroid gland.

T3 hormone is both produced within the thyroid and metabolized from T4 in peripheral tissues by means of deiodinase enzymes, marked “D” in the model below (Hoermann et al, 2022).

The angular position of this diagram seems to signify that it’s not a top-down model anymore.

Notice that the pituitary is no longer visually magnified to overwhelm the thyroid in the new, complex models shown above. It’s sandwiched between the hypothalamus and thyroid like a middle-manager in an organizational flowchart.

It’s not just an “HPT-axis” anymore, but an HPT+D-axis with “D” for deiodinases. As the model above subdivides the thyroid gland into its two functions of synthesis and metabolism, it is an “HPFD”-axis.

The deiodinases are difficult to represent visually because they are also located outside the thyroid in every tissue and organ, and inside the pituitary and hypothalamus. It would be too confusing to repeat “D” over and over visually, but that’s the reality.

After the thyroid peripheral organs work together to supply the blood with FT3, the free concentration enters all organs and tissues, including the hypothalamus and pituitary cells, where it works side by side with FT4 and co-regulates TRH and TSH secretion.

Has a given TSH level succeeded or failed?

The failure of TSH to accomplish its physiological mission is usually clear when an untreated patient’s TSH falls below 0.1 or above 10 mIU/L.

Within the range of 0.1 to 10, TSH is a hormone that can succeed or fail at any level within or outside of its population range.

What is the mission or goal of the HPT axis?

Antonio Bianco, one-time president of the American Thyroid Association, has explained that “Defending plasma T3 is a biological priority” (Abdalla & Bianco, 2014). Bianco explains in this article and later articles (Bianco et al, 2019), and Hoermann and team concur (2022), that the target hormone of the HPT axis is not TSH, but plasma T3.

“That the level of serum T3 is a main target around which serum T4 and TSH are adjusted constitutes a shift in the paradigm traditionally accepted for the function of the hypothalamus–pituitary–thyroid axis. […]

This observation challenges the dogma that locally produced T3 via D2 mixes at a prereceptor level with incoming plasma T3 in the hypothalamus and pituitary so that both, indistinctively, reduce expression of TRH and TSH. […]

Thus, experimental data support the concept that in healthy adult mice the hypothalamus–pituitary–thyroid axis is wired to preserve serum T3, except when it deliberately ‘allows’ serum T3 to drop in response to physiological or pathophysiological mechanisms, or its function is disrupted to the point that is no longer capable of reacting adequately to a fall in serum T3.”

(Abdalla & Bianco, 2014)

In other words, the hypothalamus and pituitary adjusts TSH hormone, which adjusts FT4 hormone and the T3/T4 ratio of secretion, with the main goal of defending an euthyroid level circulating FT3 supply. This is the goal of the HPT axis in “healthy” humans and in other animals like mice.

However, Bianco’s passage acknowledges there are exceptions to this general rule, because many “physiological or pathophysiological mechanisms” can permit a T3 deficiency. Mechanisms like aging, thyroid gland failure, pituitary failure, illnesses, and endocrine disruptors can get in the way of the TRH, TSH, T4 and T3 secretion rates and tissue-specific T4-T3 metabolic rates “reacting adequately to a fall in serum T3.”

These exceptions to the defense of T3 are truly puzzling, and that’s why we need thyroid endocrinologists to study the action of Free T3 hormone concentrations and intracellular T3 concentrations in the context of HPT axis health and disease — without presuming that TSH and FT4 concentrations are always in control.

“A slightly elevated TSH in subclinical hypothyroidism may accompany the successful adaptive response in some patients, but signal a failed adaption in others.”

(Hoermann et al, 2012)

If biology defends plasma T3, why would HPT axis success or failure be located at the boundaries of the TSH reference range?

  • Who has ever declared that the human thyroid gland’s hypo- or hyperfunction begins at the TSH population reference range boundaries? No reputable thyroid scientist could make such a claim.
  • Who has ever claimed that the human pituitary’s failure to compensate for any type or degree of thyroid dysfunction begins at a TSH level of 0.1, 0.4, 2.5, 4.0, or 10 mIU/L?

No one has ever proven these claims because each degree of thyroid failure (or thyroid hormone metabolic failure) requires a different level of TSH compensation.

For the sake of efficiency, health care systems want the TSH’s position in or out of range to always “flag” the existence of a thyroid pathology to the physician. But why would nature designate the 2.5th and 97th percentile of a reference range as the arbiters of successful or failed TSH adaptation in an individual?

Nature did not place TSH reference interval in the role of “flagging” an HPT axis pathology to a physician. Humans did this.

Nature did not put the “euthyroid” TSH in a huge statistical box that is far too large for any individual, and yet paradoxically, too small to contain 100% of the healthy population. Humans did this.

The reference range is defined by statistical frequency, so it’s reasonable to believe that a normal TSH will usually accompany successful adaptive response to a healthy thyroid.

However, it is also reasonable to acknowledge that organs and systems can fail in many unexpected, distorted ways. A normal TSH may occasionally coincide with a failure to stimulate a thyroid appropriately. A TSH location in the normal range is maladaptive and inappropriate when a health condition, drug, or endocrine disruptor has interfered with the pituitary’s ability to adapt TSH levels to altered thyroid function or altered peripheral metabolism.

There is no such thing as an “optimal TSH” in an individual, only a TSH that has succeeded or failed to place the individual’s FT3 and FT4 in the best place they can be given their current state of health (or illness).

How to discern thyroid dysfunction despite a normal TSH

Listen to those who have succeeded diagnosing thyroid disease within the TSH reference interval.

According to the physicians who published a case study, one brave clinical biochemist saved the life of a woman suffering multiple health problems due to 23 years of undiagnosed central hypothyroidism. They performed this heroic act simply by deciding to run a FT3 and FT4 test despite the normal TSH of 1.55 mU/L.

The authors of the 2017 article — Glyn, Harris, and Allen — reported that:

“the clinical biochemist reviewed the patient’s results and elected to add on an fT4 to her thyroid function tests.”

(Glyn et al, 2017)

Upon finally testing both Free T3 (FT3) and Free T4 (FT4) in 2015, these were her laboratory results:

  • TSH: 1.55 mU/L (0.3 – 5.5)
  • FT4: <0.3* pmol/L (12 – 22) *below the assay’s limit of detection.
  • FT3: 0.4 pmol/L (3.1 – 6.8)

How could a clinical biochemist suspect what many specialists could not?

  1. First of all, they knew the limitations of the TSH assay technology. Perhaps this individual had been involved in establishing the TSH reference interval and had to exclude blood samples with abnormal TSH-FT4-FT3 relationships that provided evidence of central hypothyroidism.
  2. Next, this person may have had some medical insight into the way hypothyroidism affects some other biochemical tests for disorders in muscles, liver, kidneys, complete blood count, electrolytes, and so on (See Franklyn & Shephard, 2000, a long list in a section called “Miscellaneous biochemical and physiological changes related to the action of thyroid hormones.”)
  3. Finally, they had instant access to run a FT4 and FT3 test without even asking the physician. Unlike a physician, they didn’t have administrative barriers in their way forbidding them from ordering FT3 and FT4 when TSH was normal, as recommended in the Choosing Wisely Canada toolkit misnamed “Understanding the Gland.” Instead of being forced to subject their clinical judgment to health care system demands, this laboratory chief was in the role of either implementing (or waiving) these institutionalized barriers.

Jonklaas and Ravzi wisely titled their 2019 article “Reference intervals in the diagnosis of thyroid dysfunction: treating patients not numbers.

They ask physicians to treat the patient, not the numbers, by listening to clinical evidence that may contradict the TSH:

“Clinicians are not able to suspect when tests are affected by these interferences and others unless the results disagree with their clinical observations.”

(Jonklaas & Ravzi, 2019)

Results that fall within the reference range do not
necessarily indicate an absence of thyroid disease
, and the complete clinical picture must be assessed.”

(Jonklaas & Ravzi, 2019)

Where does one find the evidence? Leading endocrinologists give respect back to symptoms and signs (measurable biomarkers).

“Thyroid hormone has actions in all cells of the body. Substantial perturbations in thyroid function can have manifestations in any organ system, and these effects can be well compensated and difficult to recognise due to their non-specificity, or can be obvious and classic in their presentation.”

(Jonklaas & Ravzi, 2019)

While casting doubt on TSH normality meaning “euthyroid” status, and “subclinical hypo- or hyperthyroidism” meaning the person is “not euthyroid,” Ling and team also return diagnostic weight to the evidence cholesterol and thyroid antibodies. They echoed the 2016 call by Hoermann’s team of scientists, who are leading experts in the study of the HPT axis:

“While sole reliance on TSH must therefore be scaled back, good clinical practice taking into account the full history and symptoms displayed by a patient has to be re-instituted as a primary tool.”

(Hoermann et al, 2016)

In their example above of the 76-year-old hypothyroid man, Ling, Jonklaas, Soldin and team illustrate how high-quality Free T3 and Free T4 assays can function as TSH lie-detectors when a normal TSH categorically misclassifies a patient.


It’s not uncommon for patients to hear from their doctors a sentence like “You can’t have a thyroid problem because your TSH is normal.”

The problem is with the word “can’t.” It incorrectly moves away from statistical associations and probabilities and implies that the body requires the permission of an elevated TSH before it is allowed to be in a hypothyroid state.

The best one can say is that “your normal TSH, in isolation, suggests that you may be euthyroid.”

A normal TSH has been granted too much medical authority. It is being misused to “rule out” thyroid dysfunction as either non-existent or clinically insignificant in a patient with multiple signs, symptoms, and FT3 and FT4 levels that disagree with the normal TSH.

It runs against to the purposes of “evidence-based medicine” to dismiss FT3 and FT4 and other clinical evidence if TSH is “in range,” but to grant the very same evidence weight as soon as the TSH is “out of range.”

Humans, not biology, have given TSH this authority.

No other pituitary hormone has been given this diagnostic authority:

  • No one gives normal basal ACTH levels the right to rule out adrenal gland dysfunction.
  • No one gives to normal LH and FSH levels the right to rule out primary hypogonadism (testicular hypofunction in men).

People have imagined that the scientific “evidence” says that the TSH-FT4 relationship in the healthy population is immutable and remains linear in every region, subpopulation and individual, like some sort of endocrine “law of gravity.”

Too many physicians don’t understand biology’s power to distort “normal” relationships between TSH and thyroid hormones, so they aren’t looking for exceptions to the norms. They can’t see exceptions.

Of course TSH behavior in a healthy population looks normal and orderly. But TSH behavior in a sick individual, or one with a hypothalamus-pituitary-thyroid-deiodinase disorder can be complicated by “other factors.” TSH can be whimsical, fickle, and illogically normalized by extrapituitary forces.

At times, it seems that the medical imagination has granted the pituitary TSH human-like or even god-like attributes. The anterior pituitary dangles from the hypothalamus robed in a white lab coat and stethoscope. In this comic-book version of physiology, the normal TSH means the pituitary has benign apathy toward a thyroid hormone level that does not meet cardiac demands or cerebral cortex needs. The testimony of the pituitary weighs more than a chorus of tissue-specific testimony from the heart, liver, kidney, or gastrointestinal tract. The physician then imitates this apathetic pituitary TSH “attitude” and attributes the patient’s symptoms and signs to non-thyroidal causes.

When physicians idolize TSH test results and interpret them categorically by reference intervals, it degrades the “standard of care.”

Please, someone needs teach all doctors in medical school fundamental lessons about TSH biology that they are not learning:

  1. The TSH concentration is not a mere slave to negative feedback from circulating FT3 and FT4 alone. TSH is co-regulated by many metabolic variables and is subject to many physiological and pathological interferences.
  2. The only way any TSH level can “cause” euthyroid status is through its successful feed-forward co-regulation of healthy thyroid tissue’s T3 and T4 secretion. In some people, TSH success (euthyroid status) may require TSH stimulation levels to compensate for thyroid dysfunction by rising above or falling below reference range.
  3. There is no evidence that the descriptive TSH reference range boundaries in a population act as physiological thresholds between TSH stimulation success (euthyroid status) and TSH failure (hypo- or hyperthyroid status) in an individual. “Mild” hypothyroidism and/or thyroid gland failure does not begin at the TSH reference range’s upper limit.
  4. Therefore, TSH can fail to compensate within the normal range if the thyroid gland requires a higher quantity or quality of TSH stimulation than the pituitary is able to provide. The failure of a normal TSH does not require a rare pituitary or hypothalamic disease; all that is necessary is one or many biochemical interferences.

What do undiagnosed and misdiagnosed thyroid patients need?

  • We need physicians to seek advanced thyroid re-education from the best thyroid scientists who are equipped with evidence-based TSH suspicion and don’t put TSH on a pedestal.
  • We need retired physicians and leading endocrinologists (safe from recrimination from TSH-worshipping peers) to work toward institutional removal of TSH cognitive barriers to accurate diagnosis.
  • We need clinical biochemists and ethical health care administrators who can see through the false and dangerous rhetoric of the Choosing Wisely toolkit and institute more exceptions than just “suspicion of central hypothyroidism” to the TSH-reflex or TSH-progressive testing flowchart.

Then physicians and scientists will finally have the intellectual freedom to distinguish adaptive from pathological TSH, FT3 and FT4 imbalances within the context of symptoms and other clinical evidence.


References for all articles cited in the “analyzing normal lab results” series are in a separate post.

Leave a public reply here, on our website.

This site uses Akismet to reduce spam. Learn how your comment data is processed.