Prognostic role and relationship of thyroid dysfunction and lipid profile in hospitalized heart failure patients

Abstract Background Thyroid dysfunction might have a negative impact on the prognosis of patients with heart failure (HF) and affect the lipid metabolism. The aim of our study was to investigate the prognostic role of thyroid dysfunction and its relationship with lipid profile in hospitalized HF patients. Hypothesis Thyroid dysfunction strongly correlates with prognosis of HF patients and combination with lipid profile improves the prognostic value. Methods We performed a single‐center retrospective cohort study including hospitalized HF patients between March 2009 and June 2018. Results Among enrolled 3733 patients, low fT3 (hazard ratio [HR] 1.33; 95% CI: 1.15–1.54; p < .001), elevated TSH (HR 1.37; 95% CI 1.15–1.64; p < .001), LT3S (HR 1.39; 95% CI: 1.15–1.68; p < .001), overt hyperthyroidism (HR 1.73; 95%CI: 1.00–2.98; p = .048), subclinical hypothyroidism (HR 1.43; 95%CI: 1.13–1.82; p = .003) and overt hypothyroidism (HR 1.76; 95%CI: 1.33–2.34; p < .001) independently increased the risk of composite endpoint defined as the combination of all‐cause mortality, heart transplantation, or left ventricular assist device requirement. Higher total cholesterol (HR 0.64; 95%CI: 0.49–0.83; p < .001) was still a protective factor in HF patients. When divided into four groups by fT3 and median lipid profiles, comparison of Kaplan–Meier survival curves for various groups showed good risk stratification (p < .001). Conclusion LT3S, overt hyperthyroidism, subclinical and overt hypothyroidism were independently associated with poor outcomes in HF. The combination of fT3 and lipid profile improved the prognostic value.


| INTRODUCTION
Thyroid hormones (THs) have several relevant effects on cardiac function by affecting cardiac electrophysiology and contractility, vascular resistance, and even lipid metabolism. Therefore, the occurrence of thyroid dysfunction can favor the onset and progression of several cardiovascular diseases, such as atrial or ventricular arrhythmia, dyslipidemia, atherosclerosis, hypertension, and heart failure (HF), determining a worse prognosis in terms of hospitalization, heart transplantation, and even mortality. 1 Multiple cohort studies have examined the relationship between thyroid function, and incident atrial fibrillation, HF, and coronary heart disease. In particular, subclinical hypothyroidism and hyperthyroidism were known to be associated with worsening left ventricular systolic and diastolic function, heart rate, and systemic vascular resistance, especially when thyroid-stimulating hormone (TSH) > 10 mLU/L or <0.1 mLU/L. 2,3 Furthermore, low triiodothyronine (T3) syndrome (LT3S), defined as the presence of low T3 levels with normal levels of TSH and normal free thyroxine (fT4) levels, is associated with worse outcomes in higher mortality among hospitalized HF, 4 both in patients with acute HF 5 and chronic HF. 6 THs have multiple effects on lipid metabolism, including synthesis, mobilization, and degradation. 7 It has been gradually recognized that thyroid dysfunction, including overt or subclinical hypothyroidism, is associated with higher total cholesterol (TC), low density-lipoprotein cholesterol (LDL-C) levels and triglyceride (TG).
Changes in these lipid parameters can lead to progressive lipid accumulation, plaque formation in the arteries, and deleterious effects on cardiovascular disease (CVD), the leading cause of death worldwide. 8 However, it seems that low LDL-C levels may predict a less favorable outcome in advanced HF. 9 The link between comprehensive lipid profile and thyroid function in HF is uncertain.
With complete information on thyroid function in a large cohort of HF patients, we aimed to investigate the prevalence and prognostic effect of thyroid hormone abnormality in hospitalized patients and its relationship with lipid metabolism.

| Measurement of thyroid function
The thyroid function measurement was evaluated at admission to HFCU. Twelve-hour fasting blood samples were drawn, and serumfree T3 (fT3), fT4, total T3 (TT3), total T4 (TT4), and TSH were measured by using radioimmunoassay (Immulite 2000; Siemens, Germany) in the Nuclear Medicine Department of Fuwai Hospital.

| Follow-up and endpoint
Follow-up data were obtained by reviewing the patients' hospital records, interviewing the patients via telephone, and examining the outpatient record. The composite endpoint was defined as the combination of all-cause mortality, heart transplantation, or left ventricular assist device requirement. The data was primarily obtained from death certificates, post-mortem reports, and medical records. Death caused by accidents was excluded. The median follow-up duration was 2.79 (1.00, 5.03) years.

| Statistical analysis
Descriptive statistics were used to examine the baseline characteristics of the study population at baseline. Data were described as the mean ± standard deviation for normally distributed continuous parameters, as the median (interquartile range) for skewed distributed variables, and frequencies (percentages) for categorical variables. Continuous normally distributed variables were tested with the Student's t-test, skewed variables were tested with the Mann-Whitney U test, and categorical variables were tested with chi-squared tests as appropriate. Bonferroni's correction was performed in multiple comparisons between thyroid dysfunction and euthyroidism groups, and a p < .01 was considered statistically significant. LDL-C levels of the whole population were grouped into quartiles. We also grouped LDL-C, HDL-C, TC, and TG levels of the whole population into two groups based on median value. Cox proportional hazards models were performed to assess the association between thyroid dysfunction or lipid profile and risk of death. Model 1 indicated unadjusted; Model 2 was adjusted for age, sex, and baseline BMI; Model 3 was further adjusted for systolic blood pressure (SBP), heart rate (HR), serum sodium, total bilirubin The association between thyroid function and lipid profile was determined using Spearman's correlation analysis and Kaplan-Meier analysis. Two-tailed p values < .05 were assumed as statistical significance. All analyses were performed using R 3.6.2 (R Foundation for Statistical Computing).
Significant differences were seen in BMI, systolic blood pressure, NT-proBNP, high-sensitivity C-reactive Protein (hs-CRP), NYHA class, AF, and lipid profile of patients with different thyroid status. Notably, LT3S, overt hyperthyroidism, and overt hypothyroidism were associated with lower serum lipid concentrations such as TC, TG, highdensity lipoprotein cholesterol (HDL-C) or LDL-C in HF patients.

| Survival analysis based on thyroid hormones and thyroid function
Univariate and multivariate Cox regression analysis of thyroid hormones, thyroid function, and composite endpoint are shown in

| Survival analysis based on lipid profile
In the analysis of the lipid profile shown in Table 2 Higher TG seemed to be a protective factor in HF patients both in univariate (HR 0.67; 95%CI: 0.60-0.75; p < .001) and multivariate Cox regression analysis (HR 0.77; 95%CI: 0.68-0.87; p < .001) adjusted for Model 2, however not significant in Model 3 (p = .284). Figure S1 indicated that lower TC, TG, LDL-C, and HDL-C, divided into subgroups according to median or quartiles values, showed negative outcomes in Kaplan-Meier survival curves (p < .001).

| Relationship between thyroid hormone levels and lipid profile
Analysis regarding the association between thyroid hormones and the parameters of lipid profile was shown in Table S1. fT3 is positively related with TC (ρ = .195, p < .001), TG (ρ = .238, p < .001), HDL-C (ρ = .144, p < .001) and LDL-C (ρ = .168, p < .001). Significant positive relationships were still detected between TT3, TT4, and the lipid profile mentioned above (p < .001). As shown in Figure 3, when ZHOU ET AL.  Note: Values are presented as mean ± SD or n (%). Median (25th-75th percentile) shown and Wilcoxon rank-sum test performed because these variables were non-normal distribution.
*Lipid profile were adjusted for Model 3, excluding total cholesterol. membrane and glucose and amino acid transport. 13 26 Our study showed that the combination of dyslipidemia and fT3 could significantly improve the outcome risk stratification. However, the mechanisms underlying this association still need to be elucidated.
There were a number of plausible hypotheses that may tentatively explain that such as the gastrointestinal congestion was partly associated with the development of low T3 syndrome through appetite loss and overall reduction in intestinal absorption of dietary lipids, which were also associated with malnutrition and cardiac cachexia in HF. 22 F I G U R E 3 Kaplan-Meier analysis for composite endpoint of patients subdivided into four groups based on lower limit of normal for fT3 (2.3 pg/mL) and median value of lipid profile.
Though we explored the prognostic value of thyroid dysfunction in patients with pre-existing HF requiring intense care and evaluated the association between the lipid profile and thyroid hormones in HF patients for the first time. Some limitations could not be neglected.
First, it is a single-center retrospective study, the possibility of residual confounding cannot be excluded and need to be validated in larger cohort studies. Secondly, thyroid function was assessed only once at baseline. We were unable to assess the prognostic value of transient changes in thyroid function in HF patients. Thirdly, there was heterogeneity in the cause of HF, and the medication after discharge was unavailable.

| CONCLUSION
The presence of LT3S, overt hyperthyroidism, and overt and subclinical hypothyroidism on admission for HF requiring intense care were independent poor prognostic indicators. Low fT3 and elevated TSH could serve as promising markers to predict adverse outcomes in HF patients. Low fT3 seemed to be associated with a lower level of lipid profiles in HF, and the combination of fT3 and parameters of lipid profile improves the power of risk stratification.

AUTHOR CONTRIBUTIONS
Yuhui Zhang and Jian Zhang designed and supervised the study. Ping

ACKNOWLEDGMENTS
We thank all the patients and practitioners who took part in the research. Funding statement: This work was supported by the Key