Low tri-iodothyronine syndrome improve the risk prediction of mortality in patients with acute heart failure

Background Clinical studies have suggested that low tri-iodothyronine (T3) syndrome negatively affects the clinical outcomes of patients with acute heart failure (AHF). The aim of this prospective cohort study was to evaluate the effect of low T3 syndrome in terms of prognosis and risk predictive potential in AHF. Methods Low T3 syndrome was dened by a low free T3 level (<3.1 pmol/L) accompanied by a normal thyroid-stimulating hormone level. The association between the free T3 level and mortality and the incremental risk prediction were estimated in adjusted models. Results

Conclusions Low T3 syndrome reclassi ed risk prediction for mortality beyond traditional risk factors.

Background
Heart failure is associated with high-level morbidity and mortality [1]. The cardiovascular system, particularly the heart, is a key target of thyroid hormones [2], changes in circulating hormone levels modulate cardiovascular metabolism. Low tri-iodothyronine (T3) syndrome is commonly considered to be an adaptive response to severe functional impairment [3], and has been reported in patients with heart failure. Clinical and experimental studies have shown that T3 plays fundamental roles in the modulation of cardiac contractility, ventricular remodelling and peripheral vascular resistance [4], the reduction in the activity of 5-monodeiodinase (which converts thyroid hormone [T4] to T3) is proportional to the clinical severity of heart disease [5], Changes in T4 levels are associated with signi cant alterations in heart function.
Increasing evidence suggests that low T3 syndrome is associated independently with an increased risk of adverse clinical outcomes in patients with acute heart failure (AHF) [6,8,9] and chronic heart failure [10]. Although previous studies have demonstrated the independent prognostic impact of low T3 syndrome, there is lack of study adjust for baseline N-terminal pro-brain natriuretic peptide (NT-proBNP) concentrations. Therefore, it remains uncertain whether low T3 syndrome are associated with long-term prognosis in patients with AHF, independently of renal function, NTproBNP levels and other comorbid conditions [11]. Here, we characterise the effect of low T3 syndrome in terms of prognosis and risk predictive potential in patients with AHF.

Study design
From April 2012 and August 2016, 482 consecutive patients were hospitalized for AHF in the First A liated Hospital of Nanjing Medical University. The inclusion criteria were age ≥18 years and hospitalisation because of AHF. Patients with malignant tumours, those exhibiting cognitive dysfunction or dementia and those with uncontrolled systemic disease or severe mental illness were excluded. In addition, 131 patients were excluded because they were taking amiodarone or antithyroid drugs (n = 12) or exhibited clinical hyperthyroidism (n = 9), hypothyroidism (n = 30), subclinical hyperthyroidism (n = 2) or subclinical hypothyroidism (n = 78). Our study complies with the Helsinki declaration for investigation in human patients, and the study was registered at http://www.chictr.org/ cn/ (ChiCTR-ONC -12001944).
Baseline demographic and clinical data (medical history, physical examination ndings, laboratory data, AHF aetiology and complications) were collected within 24 h of admission. The estimated glomerular ltration rate (eGFR) was calculated using the Modi cation of Diet in Renal Disease equation [12]. The left ventricular ejection fraction (LVEF) was measured by transthoracic echocardiography using the Simpson method. Chronic kidney disease was de ned as eGFR <60 mL/(min·1.73 m 2 ). Hyponatraemia was de ned as serum sodium level ≤135 mmol/L. Anaemia was de ned as haemoglobin level <130 g/L in males and <120 g/L in females. Hypoalbuminaemia was de ned as serum albumin level <34 g/L. All patients were treated in a standard manner. The thyroid function pro le was evaluated on admission or the following morning using an automated chemiluminescence immunoassay system (AutoBio 12 Co. Ltd., Zhengzhou, China). The reference intervals used were 3.10-6.80 pmol/L for serum free T3, 12.00-22.00 pmol/L for serum free T4 and 0.27-4.20 mIU/L for thyroid-stimulating hormone (TSH). Low T3 syndrome was diagnosed when the level of free T3 was <3.1 pmol/L, but the TSH level was normal.
The primary endpoint was all-cause mortality, and the secondary endpoint was cardiovascular mortality.
Prospective follow-up commenced after the measurement of thyroid function. Mortality was ascertained every 6 months by examination of hospital records or patient contact (directly or by telephone), and veri ed by viewing death certi cates held in the local disease control centre or by contact with family members.

Statistical analysis
Continuous variables are expressed as means and standard deviations (SDs) or as medians with 25 th and 75 th percentiles, and were compared using the unpaired Student's t-test or the Mann-Whitney U-test, depending on whether they were normally distributed, as revealed by the Kolmogorov-Smirnov test.
Categorical variables are presented as numbers (%) and were compared using the Pearson χ 2 test. The levels of N-terminal pro-brain natriuretic peptide (NTproBNP) were transformed logarithmically to normalise the distribution. Survival curves were drawn using the Kaplan-Meier method, and survival differences were compared using the log-rank test. Associations of low T3 syndrome with all-cause and cardiovascular mortality during follow-up were assessed using Cox proportional hazard regression analyses with adjustment sets as follows: (i) unadjusted, (ii) age and sex adjusted, (iii) adjusted for age, sex, LnNTproBNP, sodium, hypertension, eGFR, body mass index, diabetes, systolic blood pressure, hemoglobin. Furthermore, to clarify the improvement in risk prediction by low T3 syndrome over established risk factors, integrated discrimination improvement (IDI) and net reclassi cation improvement (NRI) were calculated. The interactions between low T3 syndrome and other factors (age >65 years, sex, ischaemic heart failure, hypertension, diabetes, chronic kidney disease, LVEF <40%, NTproBNP level >2,231 ng/L, anaemia, hyponatraemia, NYHA score and hypoalbuminemia) were explored using the likelihood ratio test. The linearity of risk was evaluated via restricted cubic spline [13] regression modelling of the association between the free T3 level (a continuous variable) and the risk of all-cause mortality. P values <0.05 were considered to be statistically signi cant. All statistical analyses were performed with the aid of R ver. 3.6.0 and Stata ver. 14 (STATA Corp LP).

Baseline characteristics
Of the 351 patients, 39 (11.1%) were lost to follow-up; we nally analysed 312 patients with AHF of average age 60.4±16.2 years, of whom 69.6% were male. Seventy-two (23.1%) patients exhibited low T3 syndrome, whereas the other 240 (76.9%) were euthyroid. The patients with low T3 syndrome were older, female, smoker, with more heart failure symptoms and poorer nutritional status. They also exhibited a higher NTproBNP level, a more elevated eGFR, a higher EF and a greater prevalence of electrolyte disturbance. The mean±SD free T3 levels in patients with low T3 syndrome and those who were euthyroid were 2.50±0.53 and 4.37±0.71 pmol/L, respectively (P<0.001; Table 1).
Restricted cubic spline regression was used to model continuous associations. A signi cant dosedependent association was observed between the free T3 level and the risk of all-cause and cardiovascular mortality. The free T3 level exhibited a negative linear relationship with the mortality risk ( Figure 2).

Reclassi cation
Reclassi cation of AHF patients by adding low T3 syndrome to fully adjusted model according to the occurrence of death during follow-up is summarized in table 3. The IDI after the individual inclusion of low T3 syndrome in the fully adjusted model for all-cause mortality was 2.0% (95% CI 0.1 -7.2; P = 0.030), and the NRI was 8.9% (95% CI -0.5-16.2; P = 0.232). For prediction of cardiovascular mortality, adding low T3 syndrome to the fully adjusted model increased IDI by 2.5% (95% CI 0.1-7.7, P = 0.030) and NRI by 21.3% (95%CI 7.6-50.2, P = 0.013).

Subgroup analysis
On subgroup analyses by age, sex, ischaemic heart failure, hypertension, diabetes, chronic kidney disease status, LVEF, NTproBNP level, anaemia and hyponatraemia status, NYHA score and hypoalbuminaemia status, the associations between low T3 syndrome and all-cause mortality were generally similar in all subgroups ( Figure 3).

Discussion
We found that low T3 syndrome was common in patients with AHF and was associated with increased all-cause mortality and cardiovascular mortality. A high free T3 level was associated with a lower risk of mortality. We also showed that low T3 syndrome improved risk prediction over clinical risk factors for mortality.
Several studies have shown that low T3 syndrome is associated independently with adverse clinical outcomes in patients with heart failure [6][7][8]10 . However, de nitions of low T3 syndrome and the measurement of thyroid function among different studies varied. Kannan et al. [7] found that an isolated low T3 level was associated with poor clinical outcomes in 1,365 patients with pre-existing heart failure. An isolated low T3 level was also associated with a composite endpoint including death, ventricular assistive device placement and heart transplantation (HR 2.12, 95% CI 1.65-2.72, P<0.001). In the cited work, low T3 syndrome was diagnosed when the TSH and free T4 levels were within the reference ranges and the total T3 level was below the reference range. In the cohort study of Sato et al. [8], low T3 syndrome (free T3 <2.3 pg/mL) was associated signi cantly with higher all cause-mortality in patients with heart failure (unadjusted HR 1.926, 95% CI 1.268-2.927, P = 0.002). However, in neither study were Cox regression models adjusted for BNP or NTproBNP levels, usually considered to be the most powerful prognostic factors in patients with heart failure [11]. Our nding that low T3 syndrome is indicative of a poor prognosis is consistent with those of previous studies exploring the effects of free T3 levels. Associations of low T3 syndrome with all-cause mortality were generally homogeneous across multiple clinically relevant subgroups, including patients divided by age, sex, ischaemic heart failure, hypertension, diabetes, chronic kidney disease status, LVEF, NTproBNP level, anaemia and hyponatraemia status, NYHA score and hypoalbuminaemia status. The differences between our study ndings and those of previous works are attributable to the way in which we assessed thyroid function, our de nition of low T3 syndrome and the confounders for which we adjusted in regression analyses. However, the evaluation of tissue thyroid hormone status by reference to blood data alone may be unreliable; how thyroid hormones act in target tissues such as the heart is not clear.
Our ndings provide further evidence of a consistent, dose-dependent association between the free T3 level and mortality of patients with AHF. Low T3 syndrome improved risk prediction, especially for cardiovascular mortality, beyond traditional clinical risk factors in AHF.
The mechanism by which low T3 levels cause adverse clinical outcomes may include effects on cardiac contractility, pathological cardiac remodelling and the development of peripheral vascular resistance [4,14]. Thyroid hormones critically regulate cardiac contractility by upregulating the expression of genes encoding sodium/potassium-transporting ATPases, thereby increasing expression of the myosin heavy chain (MHC) α gene and decreasing that of the MHC β gene, resulting in faster contraction [15,16]. Thyroid hormones also reduce cardiac brosis by down-regulating the expression of collagen-encoding genes and inducing metalloproteinase expression; the hormones act in an inotropic manner to upregulate the expression of β1-adrenergic receptors and modulate the activities of ionic channels [17,18]. Cardiac remodelling is a critical step during the progression of heart failure. Accumulating experimental evidence indicates that matrix metalloproteinases play important roles in the pathogenesis and progression of left ventricular dysfunction [19]. Hyroid hormones seem to signi cantly regulate metalloproteinase expression [20], and to reduce peripheral arterial resistance by acting directly on vascular smooth muscle cells to lower the mean arterial pressure. When this reduction is sensed by the kidney, the reninangiotensin-aldosterone system is activated to increase renal sodium absorption [21].
Given the roles played by thyroid hormones in cardiovascular regulation, might such hormones be useful in the treatment of heart failure? Such treatment (possibly using hormonal analogues) poses certain challenges; the available data are limited. Morkin et al. [22] demonstrated that 3,5-diiodothyropropionic acid (a T4 analogue) was tolerated well and improved cardiac performance. In patients with heart failure who received 4 weeks of therapy, the cardiac index and exercise capacity increased and the systemic vascular resistance index decreased. However, in a subsequent phase II multicentre randomised placebocontrolled study, the agent afforded no symptomatic bene t in patients with congestive heart failure [23]. In another randomised placebo-controlled study of patients with ischaemic and non-ischaemic dilated cardiomyopathy, T3 replacement therapy signi cantly improved ventricular performance; the heart rate and NTproBNP level decreased signi cantly [24]. However, none of the cited works [22-24] assessed mortality or hospital re-admission; the assumption that increases in the cardiac index and exercise capacity re ect a clinical bene t needs to be proven.
Our study has certain limitations. First, it had a single-centre observational design, and residual or unmeasured factors may have confounded the association between low T3 syndrome and mortality. However, we performed subgroup analyses to increase con dence in the internal validity of our results. Second, thyroid hormone function was measured only once; we did not measure changes in T3 levels over time. Future studies should explore whether dynamic changes in the free T3 level correlate with adverse clinical outcomes.

Conclusion
Low T3 syndrome is an independent prognostic factor predicting higher all-cause and cardiovascular mortality in patients with acute heart failure. The free T3 level is linearly, and negatively, associated with such risk. Low T3 syndrome is a clinically signi cant risk strati cation factor in patients with acute heart failure.

Declarations
Ethics approval and consent to participate Our study was approved by the Ethics Committee of the First A liated Hospital of Nanjing Medical University, and all patients gave written informed consent.

Availability of data and materials
The datasets used during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests. Data are presented as mean (SD) or median (interquartile range), or n (%). eGFR, estimated glomerular filtration rate; FT3, free tri-iodothyronine; FT4, free thyroxine; TSH, thyroid-stimulating hormone; NTproBNP, Nterminal pro-brain natriuretic peptide; NYHA, New York Heart Association functional class; LVEF, left ventricular ejection fraction; ACEI, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor blocker.   Figure 1 Kaplan-Meier survival curve for all-cause and cardiovascular mortality in patients with euthyroidism and low T3 syndrome.

Figure 2
Continuous association between free T3 levels and risk of all-cause and cardiovascular mortality. Dashed lines are 95% con dence intervals. Hazard ratios were estimated using cox regression modeling, adjusting for age, lnNTproBNP, sex, eGFR, hypertension, diabetes, body mass index, NYHA, sodium, anemia.