Functional abnormalities in the hypothalamic-pituitary-thyroid axis and in the peripheral action of thyroid hormones (THs) are common in patients with chronic kidney disease (CKD). The most consistent finding is the reduction in plasma levels of total (tT3) and free (fT3) triiodothyronine. It is common to accept that these changes are a component of the non-thyroidal illness syndrome, since the thyroid stimulating hormone (TSH) in plasma is generally within normal limits. The non-thyroidal illness syndrome and the reduced T3 levels have been interpreted as an adaptation to the process of deterioration in patients with chronic disease, with the overall objective of reducing the body's metabolism.1–4

The importance of hormonal changes in clinical outcomes of patients with CKD has only been studied in recent years. These studies have found that the decrease in T3 levels, as well as thyroxine (T4) levels, have an independent predictive value for all-cause mortality and, particularly, for cardiovascular-related death in patients with CKD in the pre-dialysis stage, in patients on haemodialysis (HD) or on peritoneal dialysis (PD).5–8

The most common cause of comorbidity and mortality in patients with CKD are cardiovascular diseases. However, little is known about the interaction between decreased TH and cardiovascular disorders in kidney patients, even though it has been shown that the primary decrease in THs has a wide range of negative effects on myocardial structure and function. Among other effects, it has been associated with decreased TH with endothelial dysfunction, as well as systolic9–11 and diastolic dysfunction, the latter as an expression of myocardial fibrosis.12

The objective of this study was to explore whether there is an association between reduced T3 levels and markers of myocardial injury—in this case with increased plasma levels of N-terminal pro-brain natriuretic peptide (NT-proBNP)—and whether this combination affects mortality and to establish whether the cardiovascular damage is a link between the decrease in THs and the risk of death in patients with CKD on dialysis.

The population under study consisted of 749 patients; of these, 400 were selected at random and invited to participate. Of these, 67 declined to participate due to problems with the sampling schedule, 21 patients were excluded due to problems coming to the hospital during the follow-up period and a further 16 patients did not attend the sampling session for the initial biochemical tests.

A total of 296 patients were included in the study. The cause of CKD was diabetic nephropathy in 50.7% of patients; unknown in 26.4%; hypertension in 6.4%; polycystic kidney in 6.1%; obstructive uropathy in 3.7%; chronic pyelonephritis in 2%; lupus in another 2% and various diseases in 2.8%. Regular medication included treatment with erythropoietin in 26%. 42% received iron supplements, mainly orally; 87% received calcium carbonate as a phosphate binder; 5% received calcitriol; 67% received at least one antihypertensive drug and 44% received two or more antihypertensive drugs; most frequently used were angiotensin-converting enzyme inhibitors. Among the 296 patients, those treated for HD (n=127) received three sessions per week, 4h per session; all were on dialysis using high-flux polysulfone filters. PD patients (continuous ambulatory PD and automated PD) (n=169) used glucose solutions.

Demographic, clinical and biochemical data are shown in Table 1. None of the subjects had a history of primary thyroid disease. Table 1 also shows values of the THs. tT3 levels were below the normal limit (tT3<0.60ng/ml) in 44.3% of patients. Low values of fT3 (fT3<2.3pg/ml) were more common (52.8%). TSH was greater than 5.0μU/ml in 72 cases (25.7%). In 21 of these patients (27.0%), it was greater than 10.0μU/ml. Patients with diabetes had lower levels of fT3 (1.87±0.66 vs. 2.01±0.52pg/ml; p<0.04) and tT3 (0.59±0.29 vs. 0.74±0.28ng/ml; p<0.001) compared to non-diabetic patients and TSH was greater than 5.0μU/ml in 32.7% of diabetic patients and only in 18.5% of non-diabetic patients (p<0.01). There was no difference between TH levels according to the dialysis modalities.

The variables relating to body composition and nutrition were out of the normal ranges in a significant number of patients. The mean BMI was in the overweight range. When BMI was assessed by gender, it was found to have low values in only 2% of patients, within normal range in 46%, and overweight or obese in 52%. Being overweight or obese was more common in women (58%) than in men (48%). The BMI and fat content measured by bioimpedance and expressed as percentage of body weight were closely linked (r=0.47; p<0.01). Mean extracellular water content expressed as a percentage of total body water was in the range of overhydration and, when it was evaluated taking into account the BMI and gender,13 overhydration was found more often in women (99%) than in men (31%). The plasma concentration of albumin was reduced (<4g/dl) in 30% of patients and serum transferrin (<200mg/dl) in 63%. In addition, CRP was high (>3.0mg/l) in 86% of cases. Concentrations of NT-proBNP and cTnI were above normal ranges in more than 50% of patients.

Table 2 shows multiple correlation analysis. Significant associations were found between the THs themselves, as well as with biochemical parameters of nutrition (albumin, transferrin), inflammation (CRP) and markers of myocardial injury (cTnI and NT-proBNP). Since several of these parameters are under the influence of THs, but are also affected by the state of hydration, the differences were analysed by classifying the patients according to whether or not they had low tT3 values and whether they were overhydrated. Fig. 1 shows these differences. It can be seen that NT-proBNP was higher in patients who were overhydrated than in those who had normal hydration, but, within each of these categories, those who had tT3 below normal limits had even higher NT-proBNP. In the case of serum albumin, the pattern was found to be reversed: overhydrated patients had lower albumin than those that showed normal hydration, but, within these categories, those that had low tT3 had even lower serum albumin. CRP was only found to be influenced by tT3: patients with low tT3 had higher CRP levels.

Fig. 1.

Serum levels of N-terminal pro-brain natriuretic peptide (top panel), albumin (middle panel) and C-reactive protein (bottom panel) when patients were classified by degree of hydration and concentration of total triiodothyronine (tT3). It is shown how the effect of tT3 is independent of the degree of hydration.

(0.16MB).

The follow-up period was 16.2±2.0 months. There were 72 deaths. The causes of death are shown in Table 3; cardiovascular deaths (sudden death, myocardial infarction, arrhythmia, heart failure, cerebrovascular disease) (n=33) accounted for more than 45% of all deaths. Table 4 shows the differences in the relevant variables among those who survived and those who did not. In addition to the expected differences in age, blood pressure and presence of diabetes, there were also significant differences in variables related to nutrition and inflammation (albumin and CRP), as well as with variables associated with myocardial injury (NT-proBNP). The difference between the two groups in values of the THs was only significant for tT3. There was no difference in the frequency or the type of medication, or the dialysis modality.

Variables which were different between those who survived and those who did not were considered in the survival analysis. Section a of Table 5 shows the univariate analysis. Risk factors for all-cause mortality were: age, systolic blood pressure, diabetes, serum albumin, glucose, CRP and NT-proBN; of the THs, only tT3 was a significant predictor. In the univariate analysis of cardiovascular mortality, all the above-mentioned variables, with the exception of age, were significant predictors.

Section b of Table 5 shows the multivariate analysis. Diabetes, CRP and tT3 were independent significant predictors of all-cause mortality. For cardiovascular mortality, the significant factors were CRP, NT-proBNP and tT3. It is important to note that tT3 maintained its significance as an independent risk factor for all-cause mortality and for cardiovascular mortality. In order to examine whether the extent of the changes in the THs affected survival, the effect of the number of hormones outside normal limits on mortality was analysed. Fig. 2 shows that the patients with two or more hormones considered to be in the range of hypothyroidism had a shorter survival.

Fig. 2.

The relationship between the number of thyroid hormones at low levels and elevated TSH is associated with survival. A higher number of abnormal hormones is associated with shorter survival.

(0.09MB).

The main findings of this study can be summarised as follows: (1) Reduced concentration of T3 in the blood is common in patients on dialysis. (2) The concentration of tT3 had, as expected, a significant relationship with biochemical markers of nutrition and inflammation. (3) A new finding was the inverse relationship between tT3 and NT-proBNP, a marker of myocardium injury, which may be the nexus between low T3 levels and cardiovascular mortality and, probably, also with all-cause mortality.

Reduced serum levels of tT3 and fT3 have previously been reported both in patients with CKD in pre-dialysis and in patients on HD and PD. The frequency of these changes has been found in between 10 and 30% of cases.1,14 Changes in T4 and TSH are less common. The direct mechanism for reducing T3 is the reduction in the enzymatic activity of 5′-monodeiodinase, responsible for converting T4 into T3, and an increase in 5′-monodeiodinase, which converts T4 to reverse T3, which is inactive in peripheral tissue.15 The conversion of T4 to T3 is much better preserved in the pituitary gland: this is because T3 and TSH remain within normal levels. However, as the state of patients’ health declines, the changes in T3 are more pronounced and, in more severe stages, there may be a progressive reduction in T4 levels and further changes in TSH.15 The proportion of patients with low T3 found in this study (44.3%) was higher than that reported in others. A small number of patients also showed a decrease in T4 levels and increased levels of TSH. It is important to mention that the patients included in this study did not have a history of thyroid disease. Elevated TSH may be linked to higher intensity of non-thyroidal illness syndrome or to changes secondary to patient improvement, as has been described in other studies,16–18 and not necessarily to primary hypothyroidism. This is why they were analysed together. The extent and the number of hormones affected were associated with worse clinical outcomes, as discussed below.

The highest proportion of TH abnormalities in this study can be explained by the larger proportion of patients with biochemical data of malnutrition and inflammation, as indicated in the correlation of T3 with serum albumin and CRP. The inverse relationship between low T3 levels and inflammation has been consistent with previous studies.19 Although the association between these factors is poorly understood, the effect of pro-inflammatory cytokines is a possible explanation. The administration of exogenous interleukin-6 to mice and humans reduces the conversion of T4 to T3 in tissue and speeds up their metabolism.20,21 Moreover, in cases of primary malnutrition, T3 levels falls significantly, probably as a mechanism to reduce energy expenditure in the body.22,23 In patients from the present study, it is difficult to separate the effect of inflammation from that of the combination of malnutrition-metabolic burnout, because both are closely associated.24 Loss of THs is another factor that has been considered to explain the low T3 levels. In patients treated with PD, up to 10% of the THs produced daily and variable quantities of thyroxine-binding globulin may be lost in the dialysate. In patients with proteinuria, there is also loss of THs in the urine.25–27 In this study, there were no differences in TH levels between HD and PD, nor with the urine volume, indicating that it had little effect.

The inverse correlation found between T3 and NT-proBNP indicates that low T3 levels may be involved in the remodelling of the myocardium, but this concept requires a more detailed analysis., Hypothyroid patients with a healthy heart have normal or low levels of NT-proBNP or BNP, and patients with hyperthyroidism and a healthy heart have elevated levels of NT-proBNP or BNP.28,29 In patients with CKD, the interaction between THs and cardiac function and remodelling occurs under different conditions: the thyroid gland is essentially normal and the myocardium has major functional and anatomical abnormalities. Two articles have reported a lower ejection fraction and a larger ventricular mass in PD and HD patients with reduced T3 levels,30,31 and another study found an association between decreased T3 and diastolic dysfunction.12 It should be noted that in these clinical conditions elevated, rather than reduced Nt-proBNP has been found, as would be expected if T3 is reduced. Despite this relationship, the reports concerning the relationship between THs and natriuretic factors—a widely recognised marker of myocardial injury—are limited in patients with CKD. additionally, abnormal remodelling also has an influence on this relationship: in a non-uraemic patient with dilated cardiomyopathy and a severe reduction in THs, contrary to what was expected, there has been found an increased myocardial atrial natriuretic peptide gene expression; this is corrected by levothyroxine replacement therapy.32 These data together suggest that the interaction between T3 and heart depends on which organ fails first. In CKD patients, volume overload, pressure overload and myocardial remodelling appear to precede thyroid disorders. In the long term, diastolic and systolic dysfunctions linked to inflammation and malnutrition lead to more significant thyroid disorders, which, in turn, lead to more severe myocardial injury, initiating a vicious cycle leading to progressive patient deterioration. Myocardial fibrosis and diastolic dysfunction are presented early on in CKD and are associated with a decrease in THs,33–35 this is similar to the observed in subclinical and experimental hypothyroidism.9,36 The role of other markers of myocardial injury, such as cTnI, may be masked by the co-linearity with NT-proBNP, although it must be mentioned that its importance in patients with CKD is not consistent.37

While diabetes, low serum albumin, high CRP and high NT-proBNP values are known as important factors associated with all-cause mortality and cardiovascular mortality,38,39 the role of the changes in THs as risk factors has only recently been recognised in kidney patients.5–7 However, the pathophysiological link between THs and mortality remains a debatable issue. In some studies, the fT3 level was involved as a risk factor for mortality,5 but, in other studies, and in accordance with the present study, tT3 was found to have greater predictive value in a panel of TH tests.7 We do not know why would tT3 be a better predictor of mortality than fT3? It has been argued that tT3, and fT3, decrease as a result of a reduction in the conversion of T4 to T3, but tT3 is also affected by a reduced number of proteins binding to THs. This reduction is induced by inflammation. This means that low tT3 levels represent a low level of hormones, while they also serve as an inflammatory marker.7 In this study, the decrease in tT3 had the greatest influence on mortality, but the number of THs in the range compatible with hypothyroidism was also associated with the progressive reduction in survival.

A causal relationship between THs and all-cause mortality and cardiovascular mortality cannot be established with the information currently available. The strong association of low T3 levels with markers of inflammation and malnutrition makes it difficult to weigh up the individual influence of each factor on mortality. Despite this, low T3 levels and high CRP levels are still independent risk factors for all-cause mortality and cardiovascular mortality, indicating that they may have an additive effect. It is also plausible that low T3 levels and high levels of inflammatory markers act in parallel and remain another chance factor. Inflammatory markers in patients with CKD have various possible origins, some of which can also be the triggering factor for non-thyroidal illness syndrome.

The data reported in this article show that low T3 levels may be an additional element, acting together with other traditional and non-traditional risk factors, in mortality. Intervention studies are necessary for a better understanding of this relationship between THs and myocardial remodelling in patients with CKD. It is important to discuss some progress that has been made in this regard in patients with dilated cardiomyopathy: acute T3 supplementation significantly reduced circulating levels of norepinephrine, aldosterone and NT-proBNP, and increased left ventricular end-diastolic volume, indicating that supplementation with THs has a significant role in the function of the myocardium in patients with chronic diseases or secondary thyroid disorders.34,40

Our study is descriptive so it has limitations, as causality cannot be established in the observed associations. In addition, the population studied has a significant number of confounding variables (such as the treatment modality, the prevalence of diabetes and the fact that it involves prevalent patients) and even the influence of other non-identified factors, as well as co-linearity between the variables that were measured, means that the lack of reproducibility of findings reported in other publications is expected. An appropriate response to this observation can only be provided with intervention studies.

With the results obtained, we can conclude that low tT3 levels are common in dialysis patients, which are associated with inflammation, malnutrition and myocardial injury. The latter may be the link between decreased THs and cardiovascular mortality. However, as some authors have indicated, interventional studies are required to stablish a cause–effect relationship between changes in THs with myocardial remodelling and mortality.

The authors express their thanks to the Coordination of Health Research of the Instituto Mexicano del Seguro Social for their sponsorship of this study (2008-785-041).

The authors have no conflicts of interest to declare.