Non-thyroidal illness syndrome and the prognosis of heart failure: a systematic review and meta-analysis

Background Heart failure (HF) is a complex and multifactorial syndrome caused by impaired heart function. The high morbidity and mortality of HF cause a heavy burden of illness worldwide. Non-thyroidal illness syndrome (NTIS) refers to aberrant serum thyroid parameters in patients without past thyroid disease. Observational studies have indicated that NTIS is associated with a higher risk of all-cause mortality in HF. This meta-analysis aimed to investigate the association between NTIS and HF prognosis. Methods Medline, Embase, Web of Science, and the Cochrane database were searched for any studies reporting an association between NTIS and HF prognosis from inception to 1 July 2022. A meta-analysis was then performed. The quality of studies was assessed using the Newcastle–Ottawa Scale. The heterogeneity of the results was assessed with I2 and Cochran's Q statistics. Sensitivity analysis and publication bias analysis were also conducted. Results A total of 626 studies were retrieved, and 18 studies were finally included in the meta-analysis. The results showed that NTIS in HF patients was significantly associated with an increased risk of all-cause mortality and major cardiovascular events (MACE), but not with in-hospital mortality. The stability of the data was validated by the sensitivity analysis. There was no indication of a publication bias in the pooled results for all-cause mortality and MACE. Conclusions This meta-analysis showed that NTIS was associated with a worse outcome in HF patients. However, the association between NTIS and in-hospital mortality of HF patients requires further investigation.


Introduction
Heart failure (HF) is a complex, multifactorial clinical syndrome characterized by dyspnea and fluid retention caused by impaired systolic or diastolic cardiac function (1,2). In 2017, there were 64.3 million cases of HF worldwide, an increase of 91.9% from 1990. Nearly half of the global increase occurred in China and India (1). According to the global burden of HF, the 1-year mortality risk is 15-30% and the 5-year mortality risk is as high as 75% in several populations (2). Additionally, the estimated annual global HF-related economic costs reached US $108 billion (1). Therefore, it is vital to identify new prognostic factors or treatment targets to reduce the HF-related burden of disease.
Non-thyroidal illness syndrome (NTIS) was first discovered in the 1970s and described as abnormal serum thyroid parameters in patients with acute illness but without prior thyroid disease. The most common abnormality is low triiodothyronine (T3) levels with normal thyroid-stimulating hormone levels (3). According to a previous observational study, NTIS may be present in up to 70% of critically ill patients and is associated with increased mortality (4). Furthermore, the results of a meta-analysis indicated that NTIS is most prevalent among HF patients. Pooled hazard ratios (HRs) from 41 studies showed that NTIS was independently associated with an elevated risk of cardiac and all-cause mortality in patients with cardiovascular disease (5). However, other researchers have found no correlation between NTIS and the HF prognosis (6,7), and supporting evidence from large-scale prospective cohort studies is scarce.
Therefore, we performed this meta-analysis to explore the association between NTIS and the HF prognosis for various endpoints and subgroups. To our knowledge, this is the first meta-analysis to systematically review these associations.

Overall design
This meta-analysis was performed according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guideline 2020 (8). We registered this study at the international prospective register of systematic reviews website (Prospero CRD: 42022345038) (https://www.crd.york.ac.uk/PROSPERO).

Search strategy and inclusion criteria
We conducted a comprehensive literature search of the Medline, Embase, Web of Science, and Cochrane databases to obtain studies focused on the association between NTIS and the HF prognosis, published from inception to 1 July 2022. The inclusion criteria were as follows: (i) HF patients were included in the study populations; (ii) cohort studies; (iii) studies investigating the association between NTIS and HF prognosis; (iv) study endpoints included allcause mortality or major cardiovascular events (MACE), or in-hospital mortality; and (v) studies published in English.
Our search strategy was therefore: (heart failure OR heart decompensation OR congestive heart failure OR HF) AND (non-thyroidal illness syndrome OR NTIS OR low T 3 syndrome OR low triiodothyronine syndrome OR euthyroid sick syndrome OR ESS). Two reviewers (XYQ and YXZ) screened the titles and abstracts of the retrieved studies. The full texts of relevant studies were further evaluated, and reviews, case reports, animal experiments, and studies without prognosis information were excluded. A third reviewer (YJC) was consulted when there was a difference of opinion between the first two reviewers.

Data extraction
The information extracted from the eligible studies included the author names, publication date, country, follow-up duration, sample size, study design, definition of NTIS, exclusion criteria, and endpoints. We also retrieved the effect size (odds ratio (OR), relative risk (RR), HR), with 95% confidence intervals (95% CIs) and adjusted covariates for multivariable analysis.

Quality assessment
We used the Newcastle-Ottawa Scale (NOS) to evaluate the quality of the eligible cohort studies (9). The NOS is a ninepoint scale that allocates points based on the selection of the cohort (0-4 points), comparability between exposure and control groups (0-2 points), and identification of outcome (0-3 points). Studies were categorized as high-, medium-, or low-quality if the NOS score was ≥7, 5-6, or ≤4 points, respectively. All the data extraction and quality assessment processes were checked by two independent reviewers (YXZ and XYQ).

Statistical analyses
The RR, OR, or HR were extracted from the original studies. In studies that only gave survival curves, we calculated the crude HR using Engauge Digitizer 4.1 software (10,11). Pooled ORs or HRs and 95% CIs were synthesized to estimate the association between NTIS and HF prognosis using a random-effect model according to the heterogeneity. The I² and Cochran's Q statistics were used to assess heterogeneity. A P-value < 0.05 for Q statistic and I² > 50% was considered substantial heterogeneity. The sensitivity analysis was performed by omitting each study to evaluate the consistency of the pooled results. Publication bias was assessed using the funnel plot and Egger's test. Obvious asymmetry of the funnel plot and X Qi et al. a P-value < 0.05 for Egger's test indicated significant publication bias. The trim and fill method would be used if publication bias existed. Subgroup analyses were conducted on disease onset of HF, NTIS definition, continent of origin of the study population, and source of HRs among different endpoints, that is, whether the HRs were reported or extracted from survival curves. A subgroup analysis according to MACE subtypes was also performed. All statistical analysis was performed using Stata software (Version 14.0). A two-sided P-value<0.05 was considered statistically significant.

Characteristics of eligible studies
The initial database search identified 656 articles. After eliminating 222 duplicate studies, we screened the titles and abstracts of the remaining records. Studies not relevant to our topic (n = 263), animal studies (n = 10), case reports (n = 21), and reviews (n = 98), were excluded. Forty-two reports were retained for further examination. We excluded four further studies that were conference abstracts or for which we had no access to the full text, five studies that were not focused on HF patients, ten that were cross-sectional studies, and five that did not have data that would enable the determination of effect size. Finally, our meta-analysis included 18 cohort studies that examined the relationship between NTIS and HF prognosis. The workflow is shown in Fig. 1.
Details of the analyzed studies are presented in Table 1. The 18 eligible studies included 2 retrospective and 16 prospective cohort studies. The sample ranged from 59 to 956. Twelve studies focused on the association between NTIS and HF all-cause mortality. Eleven studies analyzed MACE in HF while only two studies focused on in-hospital mortality in HF. Five studies reported effect size as ORs and eight as HRs. We extracted HRs from the survival curves in the last five studies. Except for one study considered medium quality (23), other included studies were considered as high quality with scores > 7 points.

Association between NTIS and all-cause mortality
Of the 12 studies that explored the association between NTIS and all-cause mortality in HF patients (Fig. 2), HRs were used as the effect size in nine, while ORs were used in three. In the synthesis result of HRs, NTIS was significantly associated with an increased risk of all-cause mortality in patients with HF (HR: 2.28, 95% CI: 2.00-1.61, P < 0.001, I 2 = 0). The pooled ORs also demonstrated a consistent result with high heterogeneity (OR: 2.51, 95% CI: 1.14-5.53, P < 0.001, I 2 = 75%).
A sensitivity analysis was performed to assess the stability of the pooled HRs/ORs by omitting each study. The association between NTIS and all-cause mortality of HF patients remained significant after the removal of each study (P < 0.05) ( Supplementary Fig. 1, see section on supplementary materials given at the end of this article). Funnel plots and Egger's test indicated no publication bias for pooled HRs and ORs (P > 0.05) ( Supplementary Fig. 3).

Association between NTIS and MACE
Eleven studies discussed the association between NTIS and MACE in HF patients (Fig. 3). The synthesis of HRs from seven studies showed that NTIS was significantly associated   with a higher risk of MACE in patients with HF (HR: 2.44, 95% CI: 1.69-3.51, P < 0.001, I 2 = 61.1%). Further, pooled ORs also confirmed a significant association between NTIS and MACE in HF patients (OR: 2.05, 95% CI: 1.32-3.19, P < 0.05, I 2 = 63.5%). The heterogeneity of the pooled HRs and ORs was significant (I 2 > 50%), so we conducted the following subgroup analyses ( Table 3).
The subgroup analysis of disease onset showed that NTIS was significantly associated with a higher risk of For the subgroup analysis of the type of MACE, the result showed that NTIS was significantly associated with an increased risk of MACE for major adverse events (MAE), readmission for HF, and cardiac death (MAE: HR: 3.70, 95% CI: 2.53-5.41, P < 0.001, I 2 = 0; readmission: HR: 2.11, 95% CI: 1.29-3.47, P = 0.003, I 2 = 0%; cardiac death: HR: 3.29, 95% CI: 2.05-5.26, P < 0.001, I 2 = 0%). However, the association was not significant in the subgroup of cardiac death and readmission for HF (HR: 1.58, 95% CI: 0.71-3.54, P = 0.265, I 2 = 75.5%). Additionally, the heterogeneity of subgroups on disease onset, definition of NTIS, continents, and source of HR decreased compared to the pooled result, which suggests the potential source of heterogeneity.
Sensitivity analysis confirmed the stability of pooled HRs and ORs ( Supplementary Fig. 2). The funnel plot and Egger's test did not reveal a tendency for publication bias (P > 0.05) (Supplementary Fig. 4).

Association between NTIS and in-hospital mortality
Two studies that reported endpoints of in-hospital mortality in HF patients were included in the metaanalysis (Fig. 4). The pooled HRs showed the association between NTIS and in-hospital mortality of HF patients was not significant (P > 0.05). We did not perform subgroup and sensitivity analysis due to the limited number of studies. The Egger's test revealed a potential publication bias (P < 0.05), but the use of the trim and fill method did not produce a reversion of the pooled result (P = 0.03, random-effect model).

Discussion
In this meta-analysis, we explored the association between NTIS and both short-and long-term outcomes of HF patients. The results indicate that NTIS is significantly associated with an increased risk of all-cause mortality, and MACE in HF patients, whereas the association with in-hospital mortality was not significant. In addition, the association between NTIS and all-cause mortality remained significant in several subgroups. However, the association between NTIS and MACE was not significant in the subgroup of AHF patients, studies from Asia and MACE defined as cardiac death and readmission to hospital.
In contrast to the earlier meta-analysis focusing on the association between NTIS and cardiovascular disease, we discussed NTIS and HF prognosis. Additionally, we analyzed the association between NTIS and multiple outcomes of HF with more studies included. In the original studies, the effect size reporting the association between NTIS and HF prognosis was not uniform. To avoid potential bias, we performed separate meta-analyses for HRs and ORs. The pooled HRs and ORs confirmed that the association between NTIS and all-cause mortality, and MACE in HF patients was significant. There was no heterogeneity in the pooled HRs for all-cause mortality.
In the subgroup analysis based on the source of HRs, the heterogeneity increased in the pooled HRs extracted from survival curves, which might be due to the differences in the ethnic and exclusion criteria of the two studies. However, after the exclusion of HRs extracted from survival curves, the result still supports the association of NTIS with HF prognosis. On the other hand, the meta-analysis of MACE revealed significant heterogeneity. However, the heterogeneity among subgroups including disease onset, definition of NTIS, source of HR, and continents of origin, was not significant, which indicated the heterogeneity source. Additionally, the association between NTIS and MACE in HF patients was not significant in several subgroups. The association was not significant in AHF patients, which may be explained by the limited number of studies, or a difference in the disease prognosis profile. A survey by the European Society of Cardiology (ESC), the ESC-HF pilot study survey, revealed that, compared with AHF, CHF had a better medium-term prognosis due to more mature therapies (12). Further studies will be needed to clarify this. Overall, this study was a comprehensive metaanalysis which confirmed the association between NTIS and HF prognosis. The results of the sensitivity analysis and lack of publication bias support the robustness of our conclusions.
Thyroid hormone (TH) plays a central role in regulating the cardiovascular system. Thyroxine is converted to the biologically active form of TH, triiodothyronine (T 3 ), by the deiodinase enzyme (13). By binding to the nuclear receptor, T 3 upregulates the expression of genes encoding for sodium/potassium-transporting ATPase, myosin heavy chain (MHC) α, and calcium ATPase protein of the sarcoplasmic reticulum (SERCA2α), which enhance systolic and diastolic cardiac function. Indirect effects of T 3 on the cardiovascular system include the activation of sodium, potassium, and calcium membrane channels, which induces the endothelial nitric oxide production, and the activation of PI3K/AKT pathways to reduce vascular resistance (13). Past studies considered NTIS an adaptive mechanism to reduce metabolic demand during HF (14). Under the pathophysiologic condition of HF, increased type 3 deiodinase activity could account for the low T 3 state (15). The persistent alterations in TH metabolism could promote HF progression through modifications in cardiac histology, morphology, and gene expression (16). A study using a rat HF model revealed that HF rats with NTIS exhibited β-MHC expression but not α-MHC expression (17). Except for the expression change of the myosin  chain, the SERCA2/phospholamban ratio also decreased in HF models (16). These alterations of cardiac genes are termed 'fetal recapitulation' and further accelerate HF progression. Physiological T 3 concentrations inhibit the process of fetal recapitulation, while low T 3 levels do not (18). Furthermore, the TRα1 (thyroid hormone receptor) isoform is thought to be a molecular switch, assisting the cardiac remodeling process in HF (18). Therefore, more recent studies suggest NTIS is a maladaptive mechanism in HF, and the bidirectional association between NTIS and HF is considered a vicious cycle (14). Our meta-analysis provides clinical evidence that NTIS is associated with HF prognosis. Further basic and clinical research is necessary to determine whether it could be a therapeutic target for HF.
Our meta-analysis had several limitations. First, there were different effect sizes reported by the included studies, which limited the number of available for synthesizing pooled ORs. Further, some HRs were extracted from survival curves. Therefore, we performed separate metaanalyses for HRs and ORs and a subgroup analysis based on the source of HRs. Second, due to a lack of clinical information, we could not further analyze the association between NTIS and the prognosis of different subtypes of HF, such as HF with reduced ejection fraction and HF with preserved ejection fraction. These deficiencies require more clinical evidence to fill the gaps.

Conclusions
In conclusion, our meta-analysis demonstrated that NTIS was significantly associated with all-cause mortality and MACE in HF patients. There was no significant association between NTIS and in-hospital mortality of HF patients. These findings suggest NTIS could be used in risk stratification for HF prognosis.

Supplementary materials
This is linked to the online version of the paper at https://doi.org/10.1530/ EC-23-0048.