This is a letter from Australian authors who have shown evidence that thyroid hormones are better tools for assessing thyroid function than TSH. It's a lengthy letter in response to a criticism by a USA guru, but it parallels our thinking to a large extent. We disagree on detail, especially the allegation that the relationship between FT4 and TSH is the same in normal and T4-treated individuals and the absence of FT3 alongside FT4, but not the general thrust. I append the letter here plus the reference it refers to:
We are grateful to Hennessey (1) for providing the opportunity to further clarify the conclusions and implications of our work (2). These are more radical than Hennessey has indicated. We believe the accumulated theoretical and empiric evidence indicates we should be discussing(a) abandoning the term ‘subclinical thyroid dysfunction’ and replacing it with ‘compensated euthyroidism’, (b) the adoption of free thyroxine (FT4) levels in place of thyroid stimulating hormone (TSH) levels as the first step/screening test for clinical assessment of thyroid function, (c) placing greater importance on the presence of abnormal thyroid hormone levels in the presence of normal TSH levels (isolated hypo/hyperthyroxinemia), and (d) in the research setting, replacing the study of ‘subclinical thyroid dysfunction’ (the study of isolated TSH abnormalities)with the study of cohorts of individuals with borderline thyroid hormone levels.
Though we noted that TSH levels are a sensitive indicator of overt thyroid dysfunction we did not mean to reassure practitioners that checking TSH levels should be the first step in clinical decision making. We believe that there are good grounds for replacing TSH levels with FT4 levels for this purpose. FT4 levels are very specific and sensitive for thyroid dysfunction, with T3 toxicosis being the one condition that might be overlooked with a normal FT4 level. TSH levels, on the other hand, do not account for hypothalamic-pituitary disorders or isolated hypo/hyperthyroxinemia. Furthermore, with the re-naming of ‘subclinical thyroid dysfunction’ to ‘compensated euthyroidism’ the specificity of abnormal TSH levels would fall significantly.The thyroid status of an individual would continue to be most accurately assessed by clinical assessment combined with two, and preferably all, of FT4, free triiodothyronine (FT3) and TSH levels.
Our work indicates that, with thyroid hormone replacement, there is no a priori reason to prefer the current practice of normalising the TSH level. In particular, the fact that we have more than one thyroid hormone is not sufficient reason. In the absence of any study it makes more sense to normalise thyroid hormone levels. There are, as Hennessey indicates, papers that suggest that the relationships between FT4, FT3 and TSH may change in the therapeutic situation (3). By contrast, Hadlow et al.demonstrated FT4/TSH relationships to be similar in treated and untreated individuals (4). Thus it may be that different levels of thyroid hormones, measured at specific times in relation to dosing, will
be associated with the clinical states, depending on the presence and nature of the thyroid therapies. When levothyroxinealone is used for thyroid replacement, the target of a FT4 level above the middle of the normal range to compensate for a relative deficiency of FT3 is an example of this proposition.
It is not possible to readily extrapolate our findings to less physiological scenarios such as thyroid replacement including the use of liothyronine (LT3). In any situation, e.g. replacement therapy with LT3, where the significance of thyroid hormone levels is not clear, TSH levels may provide useful information. It may be, however, that in such situations TSH physiology may also be altered, lessening the validity of these levels. The fact that any particular replacement therapy associated with sustained normal TSH levels is associated with mortality comparable to a background population(5) is not necessarily reassuring, as a normal background population will contain individuals with thyroid hormone levels that place them at an increased risk of death(6). With optimal thyroid replacement the risk of death might be expected to be lower than that of a background population.
TSH levels remain good tests for thyroid function and thyroid replacement. They are however only indirect measures of thyroid hormone levels, which are ‘readily available’ and ‘reflect the impact of activated thyroid hormone at a tissue level’ even better, thus allowing more precise diagnosis. The studies we included had relatively few patients on thyroid treatment and the results were similar with and without their inclusion. Physiologically, TSH levels do respond to the levels of thyroid hormones, but in the therapeutic situation just as in a ‘controlling’ situation any TSH level will correspond to a range of thyroid hormone values in the population (7).Therefore the level of TSH cannot be relied upon in either situation to indicate the thyroid state as ‘an integrated answer’ in terms of tissue response.
Only if we knew the TSH level which corresponded with mid-range thyroid hormone levels in a given individual, and we knew that TSH physiology had not been altered by the replacement therapy, would we have a suitable TSH target level for that individual, for that replacement therapy. As indicated in our paper, these questions regarding replacement therapy would be best addressed, if possible, by studies similar to ours, but restricted to individuals on the different replacement therapies.
We reiterate that our approach to the assessment of the thyroid state represents a simplification. It involves direct, rather than indirect, measurement of the thyroid state. It removes the confusion occasioned by TSH levels that do not match the levels of thyroid hormones, whilst increasing the precision of the diagnosis of the thyroid state. It helps clarify the underlying physiology bringing it into line with that of other systems.
We encourage more research, particularly regarding the relative merits of FT4 and FT3 levels, and possible interventions in individuals with borderline thyroid hormone levels. We look forward to ongoing efforts to reach consensus
Fitzgerald SP, Bean NG, Falhammar H, Tuke J. Clinical Parameters are More Likely to Be Associated with Thyroid Hormone levels than with TSH levels: A Systematic Review and Meta-Analysis. Thyroid 2020 Apr29
