Long-Term Outcome in Patients With Heart Failure Treated With Levothyroxine

An Observational Nationwide Cohort Study

Mette Nygaard Einfeldt; Anne-Marie Schjerning Olsen; Søren Lund Kristensen; Usman Khalid; Jens Faber; Christian Torp-Pedersen; Gunnar H Gislason; Christian Selmer


J Clin Endocrinol Metab. 2019;104(5):1725-1734. 

In This Article


In this large historical cohort study of patients with HF we examined the association between L-T4 substitution and mortality and cardiovascular-related endpoints. The study found that patients with HF treated with L-T4 substitution at baseline and patients with HF prescribed L-T4 during follow-up exhibited an increased risk of all-cause mortality, MACE, and cardiovascular death.

In clinical practice, it is common to adjust L-T4 treatment until patients reach and maintain a TSH level in the low end of the normal reference range, corresponding to a T4 level in the high end of the normal reference range.[29] However, recent studies of patients with high-normal T4 levels show a positive association between T4 level and increased cardiovascular mortality and morbidity.[30,31] Although we did not have information on T4, T3, and TSH levels in this study, it would not be unreasonable to speculate that many of the patients included would be receiving treatment corresponding to a high-normal level of T4 and therefore at a higher risk of cardiovascular complications.

It is known that patients with HF often develop low levels of T3 and thereby a potential intramyocardial state of hypothyroidism due to the downregulation of T3 genes that might lead to reduced intracellular metabolism in the cardiomyocytes.[7] It has been suggested that substitution with L-T4 further lowers the levels of T3 through impaired conversion of T4 into T3.[14,32] A hypothesis has been raised that patients with HF would benefit from supplementation of liothyronine (L-T3).[33,34] A study by Pingitore et al.[35] in which L-T3 was intravenously infused over 72 hours to patients with HF demonstrated a beneficial effect on neurohormonal upregulation of aldosterone, renin, and epinephrine but no beneficial effect on ejection fraction. In a recent study by Holmager et al.[36] where 13 patients with stable HF were randomly assigned to either 3 months of treatment with L-T3 (tablets) or placebo, no increase in ejection fraction and no changes in neurohormones were found. On the other hand, patients in the study by Holmager et al. treated with L-T3 did not experience tachycardia. These findings support neither of the hypotheses that a low T3 level has a protective or harmful effect on cardiomyocytes.[7,8]

Low thyroid function in patients with HF has been associated with worsening in clinical prognosis of HF[37,38] and increased risk of hospitalization.[17,39,40] Because of the uncertainty about whether L-T4 has a beneficial effect on mortality, substitution treatment with L-T4 in patients with HF is still subject to debate, with data limited to a few small intervention studies.[15–17] Results from the prospective Cardiovascular Health Study of 3044 adults aged ≥65 years found that in the 46 patients with subclinical hypothyroidism and TSH >10 mIU/L, the patients treated with L-T4 during follow-up had a higher risk of developing HF than those not treated with L-T4.[41] However, our findings do not support the hypothesis that patients with known HF benefit from L-T4 substitution. One explanation could be that patients who are substituted with L-T4 do not tolerate thyroid hormone replacement therapy, because of either the unclassified severity of their underlying HF condition or potential overtreatment.[42] Another possible explanation is that patients treated with L-T4 may have an overall inferior medical condition than those not treated and that these unmeasured confounders bias our findings. In particular, some patients with HF treated with L-T4 may have a mild "low-T3 syndrome" with a simultaneous increased TSH and therefore may have a worse clinical condition than those not treated with L-T4.[43]Another interpretation of our results is that the elevated risk of death and adverse cardiovascular death is due to thyroid dysfunction rather than L-T4 treatment. Because of the limitations inherent to the observational study design, this interpretation cannot be refuted, although in a previous study by our group, no risk of increased mortality was found in patients with low thyroid function substituted with L-T4 compared with euthyroid patients.[44]

The current study also showed an increased risk of MI in patients who were treated with L-T4 at baseline but a significantly decreased risk of MI in the patients who initiated L-T4 during follow-up. One hypothesis could be that patients with recent MI and a subsequent diagnosis of hypothyroidism during follow-up are substituted more carefully to avoid overtreatment, but this possibility remains speculative.

Strengths and Limitations

The main strength of this study was the nationwide cohort of "real-world" patients with first-time admission for HF. Another strength of this study is the size of the study population and the extensive information on comorbidity and pharmacotherapy, which included prescription medication. The HF diagnosis in the Danish registers has a high specificity of 99%, rendering the registered discharge diagnosis suitable for selecting patients for the study because the diagnosis accurately identifies patients with clinical HF.[23,45]

Limitations include the lack of clinical information, such as echocardiography, left ventricular ejection fraction, plasma natriuretic peptide, New York Heart Association functional classification, blood pressure, serum lipid, body mass index, smoking status, or the specific cause of thyroid dysfunction, because we do not hold information about radioiodine treatment or thyroid autoantibodies as well as TSH, T3, and T4 levels. T3 measurements in particular would have been valuable as a predictor for the severity of HF, because a study by Iervasi et al.[43] found "low-T3 syndrome" to be a strong predictor of death in cardiac patients.

Our findings could be influenced by confounding by indication because we do not know what caused physicians to prescribe L-T4 treatment to some patients and not to others. One could speculate that physicians might be more likely to prescribe L-T4 to patients with severe HF, which could explain the increased risk observed in patients treated with L-T4.

Also, it is generally not possible to draw conclusions on causality based on observational data alone. Specifically, it is not possible with the available data to examine whether the significant increased mortality in L-T4–treated patients is caused by the thyroid dysfunction or the impact of thyroid hormone substitution on HF.